Mercurial > repos > rhpvorderman > shm_csr
annotate baseline/Baseline_Functions.r @ 4:c8f02bce10d0 draft
"planemo upload commit 05326cc2233eb0fa8beaf4198bd7c1519f2a0bf9"
| author | rhpvorderman |
|---|---|
| date | Mon, 15 Nov 2021 14:04:10 +0000 |
| parents | 64d74ba01a7c |
| children |
| rev | line source |
|---|---|
|
0
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1 ######################################################################################### |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
2 # License Agreement |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
3 # |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
4 # THIS WORK IS PROVIDED UNDER THE TERMS OF THIS CREATIVE COMMONS PUBLIC LICENSE |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
5 # ("CCPL" OR "LICENSE"). THE WORK IS PROTECTED BY COPYRIGHT AND/OR OTHER |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
6 # APPLICABLE LAW. ANY USE OF THE WORK OTHER THAN AS AUTHORIZED UNDER THIS LICENSE |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
7 # OR COPYRIGHT LAW IS PROHIBITED. |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
8 # |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
9 # BY EXERCISING ANY RIGHTS TO THE WORK PROVIDED HERE, YOU ACCEPT AND AGREE TO BE |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
10 # BOUND BY THE TERMS OF THIS LICENSE. TO THE EXTENT THIS LICENSE MAY BE CONSIDERED |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
11 # TO BE A CONTRACT, THE LICENSOR GRANTS YOU THE RIGHTS CONTAINED HERE IN |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
12 # CONSIDERATION OF YOUR ACCEPTANCE OF SUCH TERMS AND CONDITIONS. |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
13 # |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
14 # BASELIne: Bayesian Estimation of Antigen-Driven Selection in Immunoglobulin Sequences |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
15 # Coded by: Mohamed Uduman & Gur Yaari |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
16 # Copyright 2012 Kleinstein Lab |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
17 # Version: 1.3 (01/23/2014) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
18 ######################################################################################### |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
19 |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
20 # Global variables |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
21 |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
22 FILTER_BY_MUTATIONS = 1000 |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
23 |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
24 # Nucleotides |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
25 NUCLEOTIDES = c("A","C","G","T") |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
26 |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
27 # Amino Acids |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
28 AMINO_ACIDS <- c("F", "F", "L", "L", "S", "S", "S", "S", "Y", "Y", "*", "*", "C", "C", "*", "W", "L", "L", "L", "L", "P", "P", "P", "P", "H", "H", "Q", "Q", "R", "R", "R", "R", "I", "I", "I", "M", "T", "T", "T", "T", "N", "N", "K", "K", "S", "S", "R", "R", "V", "V", "V", "V", "A", "A", "A", "A", "D", "D", "E", "E", "G", "G", "G", "G") |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
29 names(AMINO_ACIDS) <- c("TTT", "TTC", "TTA", "TTG", "TCT", "TCC", "TCA", "TCG", "TAT", "TAC", "TAA", "TAG", "TGT", "TGC", "TGA", "TGG", "CTT", "CTC", "CTA", "CTG", "CCT", "CCC", "CCA", "CCG", "CAT", "CAC", "CAA", "CAG", "CGT", "CGC", "CGA", "CGG", "ATT", "ATC", "ATA", "ATG", "ACT", "ACC", "ACA", "ACG", "AAT", "AAC", "AAA", "AAG", "AGT", "AGC", "AGA", "AGG", "GTT", "GTC", "GTA", "GTG", "GCT", "GCC", "GCA", "GCG", "GAT", "GAC", "GAA", "GAG", "GGT", "GGC", "GGA", "GGG") |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
30 names(AMINO_ACIDS) <- names(AMINO_ACIDS) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
31 |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
32 #Amino Acid Traits |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
33 #"*" "A" "C" "D" "E" "F" "G" "H" "I" "K" "L" "M" "N" "P" "Q" "R" "S" "T" "V" "W" "Y" |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
34 #B = "Hydrophobic/Burried" N = "Intermediate/Neutral" S="Hydrophilic/Surface") |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
35 TRAITS_AMINO_ACIDS_CHOTHIA98 <- c("*","N","B","S","S","B","N","N","B","S","B","B","S","N","S","S","N","N","B","B","N") |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
36 names(TRAITS_AMINO_ACIDS_CHOTHIA98) <- sort(unique(AMINO_ACIDS)) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
37 TRAITS_AMINO_ACIDS <- array(NA,21) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
38 |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
39 # Codon Table |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
40 CODON_TABLE <- as.data.frame(matrix(NA,ncol=64,nrow=12)) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
41 |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
42 # Substitution Model: Smith DS et al. 1996 |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
43 substitution_Literature_Mouse <- matrix(c(0, 0.156222928, 0.601501588, 0.242275484, 0.172506739, 0, 0.241239892, 0.586253369, 0.54636291, 0.255795364, 0, 0.197841727, 0.290240811, 0.467680608, 0.24207858, 0),nrow=4,byrow=T,dimnames=list(NUCLEOTIDES,NUCLEOTIDES)) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
44 substitution_Flu_Human <- matrix(c(0,0.2795596,0.5026927,0.2177477,0.1693210,0,0.3264723,0.5042067,0.4983549,0.3328321,0,0.1688130,0.2021079,0.4696077,0.3282844,0),4,4,byrow=T,dimnames=list(NUCLEOTIDES,NUCLEOTIDES)) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
45 substitution_Flu25_Human <- matrix(c(0,0.2580641,0.5163685,0.2255674,0.1541125,0,0.3210224,0.5248651,0.5239281,0.3101292,0,0.1659427,0.1997207,0.4579444,0.3423350,0),4,4,byrow=T,dimnames=list(NUCLEOTIDES,NUCLEOTIDES)) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
46 load("FiveS_Substitution.RData") |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
47 |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
48 # Mutability Models: Shapiro GS et al. 2002 |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
49 triMutability_Literature_Human <- matrix(c(0.24, 1.2, 0.96, 0.43, 2.14, 2, 1.11, 1.9, 0.85, 1.83, 2.36, 1.31, 0.82, 0.52, 0.89, 1.33, 1.4, 0.82, 1.83, 0.73, 1.83, 1.62, 1.53, 0.57, 0.92, 0.42, 0.42, 1.47, 3.44, 2.58, 1.18, 0.47, 0.39, 1.12, 1.8, 0.68, 0.47, 2.19, 2.35, 2.19, 1.05, 1.84, 1.26, 0.28, 0.98, 2.37, 0.66, 1.58, 0.67, 0.92, 1.76, 0.83, 0.97, 0.56, 0.75, 0.62, 2.26, 0.62, 0.74, 1.11, 1.16, 0.61, 0.88, 0.67, 0.37, 0.07, 1.08, 0.46, 0.31, 0.94, 0.62, 0.57, 0.29, NA, 1.44, 0.46, 0.69, 0.57, 0.24, 0.37, 1.1, 0.99, 1.39, 0.6, 2.26, 1.24, 1.36, 0.52, 0.33, 0.26, 1.25, 0.37, 0.58, 1.03, 1.2, 0.34, 0.49, 0.33, 2.62, 0.16, 0.4, 0.16, 0.35, 0.75, 1.85, 0.94, 1.61, 0.85, 2.09, 1.39, 0.3, 0.52, 1.33, 0.29, 0.51, 0.26, 0.51, 3.83, 2.01, 0.71, 0.58, 0.62, 1.07, 0.28, 1.2, 0.74, 0.25, 0.59, 1.09, 0.91, 1.36, 0.45, 2.89, 1.27, 3.7, 0.69, 0.28, 0.41, 1.17, 0.56, 0.93, 3.41, 1, 1, NA, 5.9, 0.74, 2.51, 2.24, 2.24, 1.95, 3.32, 2.34, 1.3, 2.3, 1, 0.66, 0.73, 0.93, 0.41, 0.65, 0.89, 0.65, 0.32, NA, 0.43, 0.85, 0.43, 0.31, 0.31, 0.23, 0.29, 0.57, 0.71, 0.48, 0.44, 0.76, 0.51, 1.7, 0.85, 0.74, 2.23, 2.08, 1.16, 0.51, 0.51, 1, 0.5, NA, NA, 0.71, 2.14), nrow=64,byrow=T) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
50 triMutability_Literature_Mouse <- matrix(c(1.31, 1.35, 1.42, 1.18, 2.02, 2.02, 1.02, 1.61, 1.99, 1.42, 2.01, 1.03, 2.02, 0.97, 0.53, 0.71, 1.19, 0.83, 0.96, 0.96, 0, 1.7, 2.22, 0.59, 1.24, 1.07, 0.51, 1.68, 3.36, 3.36, 1.14, 0.29, 0.33, 0.9, 1.11, 0.63, 1.08, 2.07, 2.27, 1.74, 0.22, 1.19, 2.37, 1.15, 1.15, 1.56, 0.81, 0.34, 0.87, 0.79, 2.13, 0.49, 0.85, 0.97, 0.36, 0.82, 0.66, 0.63, 1.15, 0.94, 0.85, 0.25, 0.93, 1.19, 0.4, 0.2, 0.44, 0.44, 0.88, 1.06, 0.77, 0.39, 0, 0, 0, 0, 0, 0, 0.43, 0.43, 0.86, 0.59, 0.59, 0, 1.18, 0.86, 2.9, 1.66, 0.4, 0.2, 1.54, 0.43, 0.69, 1.71, 0.68, 0.55, 0.91, 0.7, 1.71, 0.09, 0.27, 0.63, 0.2, 0.45, 1.01, 1.63, 0.96, 1.48, 2.18, 1.2, 1.31, 0.66, 2.13, 0.49, 0, 0, 0, 2.97, 2.8, 0.79, 0.4, 0.5, 0.4, 0.11, 1.68, 0.42, 0.13, 0.44, 0.93, 0.71, 1.11, 1.19, 2.71, 1.08, 3.43, 0.4, 0.67, 0.47, 1.02, 0.14, 1.56, 1.98, 0.53, 0.33, 0.63, 2.06, 1.77, 1.46, 3.74, 2.93, 2.1, 2.18, 0.78, 0.73, 2.93, 0.63, 0.57, 0.17, 0.85, 0.52, 0.31, 0.31, 0, 0, 0.51, 0.29, 0.83, 0.54, 0.28, 0.47, 0.9, 0.99, 1.24, 2.47, 0.73, 0.23, 1.13, 0.24, 2.12, 0.24, 0.33, 0.83, 1.41, 0.62, 0.28, 0.35, 0.77, 0.17, 0.72, 0.58, 0.45, 0.41), nrow=64,byrow=T) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
51 triMutability_Names <- c("AAA", "AAC", "AAG", "AAT", "ACA", "ACC", "ACG", "ACT", "AGA", "AGC", "AGG", "AGT", "ATA", "ATC", "ATG", "ATT", "CAA", "CAC", "CAG", "CAT", "CCA", "CCC", "CCG", "CCT", "CGA", "CGC", "CGG", "CGT", "CTA", "CTC", "CTG", "CTT", "GAA", "GAC", "GAG", "GAT", "GCA", "GCC", "GCG", "GCT", "GGA", "GGC", "GGG", "GGT", "GTA", "GTC", "GTG", "GTT", "TAA", "TAC", "TAG", "TAT", "TCA", "TCC", "TCG", "TCT", "TGA", "TGC", "TGG", "TGT", "TTA", "TTC", "TTG", "TTT") |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
52 load("FiveS_Mutability.RData") |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
53 |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
54 # Functions |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
55 |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
56 # Translate codon to amino acid |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
57 translateCodonToAminoAcid<-function(Codon){ |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
58 return(AMINO_ACIDS[Codon]) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
59 } |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
60 |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
61 # Translate amino acid to trait change |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
62 translateAminoAcidToTraitChange<-function(AminoAcid){ |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
63 return(TRAITS_AMINO_ACIDS[AminoAcid]) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
64 } |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
65 |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
66 # Initialize Amino Acid Trait Changes |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
67 initializeTraitChange <- function(traitChangeModel=1,species=1,traitChangeFileName=NULL){ |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
68 if(!is.null(traitChangeFileName)){ |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
69 tryCatch( |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
70 traitChange <- read.delim(traitChangeFileName,sep="\t",header=T) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
71 , error = function(ex){ |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
72 cat("Error|Error reading trait changes. Please check file name/path and format.\n") |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
73 q() |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
74 } |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
75 ) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
76 }else{ |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
77 traitChange <- TRAITS_AMINO_ACIDS_CHOTHIA98 |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
78 } |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
79 TRAITS_AMINO_ACIDS <<- traitChange |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
80 } |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
81 |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
82 # Read in formatted nucleotide substitution matrix |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
83 initializeSubstitutionMatrix <- function(substitutionModel,species,subsMatFileName=NULL){ |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
84 if(!is.null(subsMatFileName)){ |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
85 tryCatch( |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
86 subsMat <- read.delim(subsMatFileName,sep="\t",header=T) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
87 , error = function(ex){ |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
88 cat("Error|Error reading substitution matrix. Please check file name/path and format.\n") |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
89 q() |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
90 } |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
91 ) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
92 if(sum(apply(subsMat,1,sum)==1)!=4) subsMat = t(apply(subsMat,1,function(x)x/sum(x))) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
93 }else{ |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
94 if(substitutionModel==1)subsMat <- substitution_Literature_Mouse |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
95 if(substitutionModel==2)subsMat <- substitution_Flu_Human |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
96 if(substitutionModel==3)subsMat <- substitution_Flu25_Human |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
97 |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
98 } |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
99 |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
100 if(substitutionModel==0){ |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
101 subsMat <- matrix(1,4,4) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
102 subsMat[,] = 1/3 |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
103 subsMat[1,1] = 0 |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
104 subsMat[2,2] = 0 |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
105 subsMat[3,3] = 0 |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
106 subsMat[4,4] = 0 |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
107 } |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
108 |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
109 |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
110 NUCLEOTIDESN = c(NUCLEOTIDES,"N", "-") |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
111 if(substitutionModel==5){ |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
112 subsMat <- FiveS_Substitution |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
113 return(subsMat) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
114 }else{ |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
115 subsMat <- rbind(subsMat,rep(NA,4),rep(NA,4)) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
116 return( matrix(data.matrix(subsMat),6,4,dimnames=list(NUCLEOTIDESN,NUCLEOTIDES) ) ) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
117 } |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
118 } |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
119 |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
120 |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
121 # Read in formatted Mutability file |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
122 initializeMutabilityMatrix <- function(mutabilityModel=1, species=1,mutabilityMatFileName=NULL){ |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
123 if(!is.null(mutabilityMatFileName)){ |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
124 tryCatch( |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
125 mutabilityMat <- read.delim(mutabilityMatFileName,sep="\t",header=T) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
126 , error = function(ex){ |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
127 cat("Error|Error reading mutability matrix. Please check file name/path and format.\n") |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
128 q() |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
129 } |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
130 ) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
131 }else{ |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
132 mutabilityMat <- triMutability_Literature_Human |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
133 if(species==2) mutabilityMat <- triMutability_Literature_Mouse |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
134 } |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
135 |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
136 if(mutabilityModel==0){ mutabilityMat <- matrix(1,64,3)} |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
137 |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
138 if(mutabilityModel==5){ |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
139 mutabilityMat <- FiveS_Mutability |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
140 return(mutabilityMat) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
141 }else{ |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
142 return( matrix( data.matrix(mutabilityMat), 64, 3, dimnames=list(triMutability_Names,1:3)) ) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
143 } |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
144 } |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
145 |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
146 # Read FASTA file formats |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
147 # Modified from read.fasta from the seqinR package |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
148 baseline.read.fasta <- |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
149 function (file = system.file("sequences/sample.fasta", package = "seqinr"), |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
150 seqtype = c("DNA", "AA"), as.string = FALSE, forceDNAtolower = TRUE, |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
151 set.attributes = TRUE, legacy.mode = TRUE, seqonly = FALSE, |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
152 strip.desc = FALSE, sizeof.longlong = .Machine$sizeof.longlong, |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
153 endian = .Platform$endian, apply.mask = TRUE) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
154 { |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
155 seqtype <- match.arg(seqtype) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
156 |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
157 lines <- readLines(file) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
158 |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
159 if (legacy.mode) { |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
160 comments <- grep("^;", lines) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
161 if (length(comments) > 0) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
162 lines <- lines[-comments] |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
163 } |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
164 |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
165 |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
166 ind_groups<-which(substr(lines, 1L, 3L) == ">>>") |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
167 lines_mod<-lines |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
168 |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
169 if(!length(ind_groups)){ |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
170 lines_mod<-c(">>>All sequences combined",lines) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
171 } |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
172 |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
173 ind_groups<-which(substr(lines_mod, 1L, 3L) == ">>>") |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
174 |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
175 lines <- array("BLA",dim=(length(ind_groups)+length(lines_mod))) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
176 id<-sapply(1:length(ind_groups),function(i)ind_groups[i]+i-1)+1 |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
177 lines[id] <- "THIS IS A FAKE SEQUENCE" |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
178 lines[-id] <- lines_mod |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
179 rm(lines_mod) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
180 |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
181 ind <- which(substr(lines, 1L, 1L) == ">") |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
182 nseq <- length(ind) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
183 if (nseq == 0) { |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
184 stop("no line starting with a > character found") |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
185 } |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
186 start <- ind + 1 |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
187 end <- ind - 1 |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
188 |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
189 while( any(which(ind%in%end)) ){ |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
190 ind=ind[-which(ind%in%end)] |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
191 nseq <- length(ind) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
192 if (nseq == 0) { |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
193 stop("no line starting with a > character found") |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
194 } |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
195 start <- ind + 1 |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
196 end <- ind - 1 |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
197 } |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
198 |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
199 end <- c(end[-1], length(lines)) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
200 sequences <- lapply(seq_len(nseq), function(i) paste(lines[start[i]:end[i]], collapse = "")) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
201 if (seqonly) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
202 return(sequences) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
203 nomseq <- lapply(seq_len(nseq), function(i) { |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
204 |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
205 #firstword <- strsplit(lines[ind[i]], " ")[[1]][1] |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
206 substr(lines[ind[i]], 2, nchar(lines[ind[i]])) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
207 |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
208 }) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
209 if (seqtype == "DNA") { |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
210 if (forceDNAtolower) { |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
211 sequences <- as.list(tolower(chartr(".","-",sequences))) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
212 }else{ |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
213 sequences <- as.list(toupper(chartr(".","-",sequences))) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
214 } |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
215 } |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
216 if (as.string == FALSE) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
217 sequences <- lapply(sequences, s2c) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
218 if (set.attributes) { |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
219 for (i in seq_len(nseq)) { |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
220 Annot <- lines[ind[i]] |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
221 if (strip.desc) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
222 Annot <- substr(Annot, 2L, nchar(Annot)) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
223 attributes(sequences[[i]]) <- list(name = nomseq[[i]], |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
224 Annot = Annot, class = switch(seqtype, AA = "SeqFastaAA", |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
225 DNA = "SeqFastadna")) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
226 } |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
227 } |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
228 names(sequences) <- nomseq |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
229 return(sequences) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
230 } |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
231 |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
232 |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
233 # Replaces non FASTA characters in input files with N |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
234 replaceNonFASTAChars <-function(inSeq="ACGTN-AApA"){ |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
235 gsub('[^ACGTNacgt[:punct:]-[:punct:].]','N',inSeq,perl=TRUE) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
236 } |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
237 |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
238 # Find the germlines in the FASTA list |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
239 germlinesInFile <- function(seqIDs){ |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
240 firstChar = sapply(seqIDs,function(x){substr(x,1,1)}) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
241 secondChar = sapply(seqIDs,function(x){substr(x,2,2)}) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
242 return(firstChar==">" & secondChar!=">") |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
243 } |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
244 |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
245 # Find the groups in the FASTA list |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
246 groupsInFile <- function(seqIDs){ |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
247 sapply(seqIDs,function(x){substr(x,1,2)})==">>" |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
248 } |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
249 |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
250 # In the process of finding germlines/groups, expand from the start to end of the group |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
251 expandTillNext <- function(vecPosToID){ |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
252 IDs = names(vecPosToID) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
253 posOfInterests = which(vecPosToID) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
254 |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
255 expandedID = rep(NA,length(IDs)) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
256 expandedIDNames = gsub(">","",IDs[posOfInterests]) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
257 startIndexes = c(1,posOfInterests[-1]) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
258 stopIndexes = c(posOfInterests[-1]-1,length(IDs)) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
259 expandedID = unlist(sapply(1:length(startIndexes),function(i){ |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
260 rep(i,stopIndexes[i]-startIndexes[i]+1) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
261 })) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
262 names(expandedID) = unlist(sapply(1:length(startIndexes),function(i){ |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
263 rep(expandedIDNames[i],stopIndexes[i]-startIndexes[i]+1) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
264 })) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
265 return(expandedID) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
266 } |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
267 |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
268 # Process FASTA (list) to return a matrix[input, germline) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
269 processInputAdvanced <- function(inputFASTA){ |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
270 |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
271 seqIDs = names(inputFASTA) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
272 numbSeqs = length(seqIDs) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
273 posGermlines1 = germlinesInFile(seqIDs) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
274 numbGermlines = sum(posGermlines1) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
275 posGroups1 = groupsInFile(seqIDs) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
276 numbGroups = sum(posGroups1) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
277 consDef = NA |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
278 |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
279 if(numbGermlines==0){ |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
280 posGermlines = 2 |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
281 numbGermlines = 1 |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
282 } |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
283 |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
284 glPositionsSum = cumsum(posGermlines1) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
285 glPositions = table(glPositionsSum) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
286 #Find the position of the conservation row |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
287 consDefPos = as.numeric(names(glPositions[names(glPositions)!=0 & glPositions==1]))+1 |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
288 if( length(consDefPos)> 0 ){ |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
289 consDefID = match(consDefPos, glPositionsSum) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
290 #The coservation rows need to be pulled out and stores seperately |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
291 consDef = inputFASTA[consDefID] |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
292 inputFASTA = inputFASTA[-consDefID] |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
293 |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
294 seqIDs = names(inputFASTA) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
295 numbSeqs = length(seqIDs) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
296 posGermlines1 = germlinesInFile(seqIDs) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
297 numbGermlines = sum(posGermlines1) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
298 posGroups1 = groupsInFile(seqIDs) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
299 numbGroups = sum(posGroups1) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
300 if(numbGermlines==0){ |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
301 posGermlines = 2 |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
302 numbGermlines = 1 |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
303 } |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
304 } |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
305 |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
306 posGroups <- expandTillNext(posGroups1) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
307 posGermlines <- expandTillNext(posGermlines1) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
308 posGermlines[posGroups1] = 0 |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
309 names(posGermlines)[posGroups1] = names(posGroups)[posGroups1] |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
310 posInput = rep(TRUE,numbSeqs) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
311 posInput[posGroups1 | posGermlines1] = FALSE |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
312 |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
313 matInput = matrix(NA, nrow=sum(posInput), ncol=2) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
314 rownames(matInput) = seqIDs[posInput] |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
315 colnames(matInput) = c("Input","Germline") |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
316 |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
317 vecInputFASTA = unlist(inputFASTA) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
318 matInput[,1] = vecInputFASTA[posInput] |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
319 matInput[,2] = vecInputFASTA[ which( names(inputFASTA)%in%paste(">",names(posGermlines)[posInput],sep="") )[ posGermlines[posInput]] ] |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
320 |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
321 germlines = posGermlines[posInput] |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
322 groups = posGroups[posInput] |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
323 |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
324 return( list("matInput"=matInput, "germlines"=germlines, "groups"=groups, "conservationDefinition"=consDef )) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
325 } |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
326 |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
327 |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
328 # Replace leading and trailing dashes in the sequence |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
329 replaceLeadingTrailingDashes <- function(x,readEnd){ |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
330 iiGap = unlist(gregexpr("-",x[1])) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
331 ggGap = unlist(gregexpr("-",x[2])) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
332 #posToChange = intersect(iiGap,ggGap) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
333 |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
334 |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
335 seqIn = replaceLeadingTrailingDashesHelper(x[1]) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
336 seqGL = replaceLeadingTrailingDashesHelper(x[2]) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
337 seqTemplate = rep('N',readEnd) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
338 seqIn <- c(seqIn,seqTemplate[(length(seqIn)+1):readEnd]) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
339 seqGL <- c(seqGL,seqTemplate[(length(seqGL)+1):readEnd]) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
340 # if(posToChange!=-1){ |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
341 # seqIn[posToChange] = "-" |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
342 # seqGL[posToChange] = "-" |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
343 # } |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
344 |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
345 seqIn = c2s(seqIn[1:readEnd]) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
346 seqGL = c2s(seqGL[1:readEnd]) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
347 |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
348 lenGL = nchar(seqGL) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
349 if(lenGL<readEnd){ |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
350 seqGL = paste(seqGL,c2s(rep("N",readEnd-lenGL)),sep="") |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
351 } |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
352 |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
353 lenInput = nchar(seqIn) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
354 if(lenInput<readEnd){ |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
355 seqIn = paste(seqIn,c2s(rep("N",readEnd-lenInput)),sep="") |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
356 } |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
357 return( c(seqIn,seqGL) ) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
358 } |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
359 |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
360 replaceLeadingTrailingDashesHelper <- function(x){ |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
361 grepResults = gregexpr("-*",x) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
362 grepResultsPos = unlist(grepResults) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
363 grepResultsLen = attr(grepResults[[1]],"match.length") |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
364 #print(paste("x = '", x, "'", sep="")) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
365 x = s2c(x) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
366 if(x[1]=="-"){ |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
367 x[1:grepResultsLen[1]] = "N" |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
368 } |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
369 if(x[length(x)]=="-"){ |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
370 x[(length(x)-grepResultsLen[length(grepResultsLen)]+1):length(x)] = "N" |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
371 } |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
372 return(x) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
373 } |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
374 |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
375 |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
376 |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
377 |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
378 # Check sequences for indels |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
379 checkForInDels <- function(matInputP){ |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
380 insPos <- checkInsertion(matInputP) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
381 delPos <- checkDeletions(matInputP) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
382 return(list("Insertions"=insPos, "Deletions"=delPos)) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
383 } |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
384 |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
385 # Check sequences for insertions |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
386 checkInsertion <- function(matInputP){ |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
387 insertionCheck = apply( matInputP,1, function(x){ |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
388 inputGaps <- as.vector( gregexpr("-",x[1])[[1]] ) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
389 glGaps <- as.vector( gregexpr("-",x[2])[[1]] ) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
390 return( is.finite( match(FALSE, glGaps%in%inputGaps ) ) ) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
391 }) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
392 return(as.vector(insertionCheck)) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
393 } |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
394 # Fix inserstions |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
395 fixInsertions <- function(matInputP){ |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
396 insPos <- checkInsertion(matInputP) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
397 sapply((1:nrow(matInputP))[insPos],function(rowIndex){ |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
398 x <- matInputP[rowIndex,] |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
399 inputGaps <- gregexpr("-",x[1])[[1]] |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
400 glGaps <- gregexpr("-",x[2])[[1]] |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
401 posInsertions <- glGaps[!(glGaps%in%inputGaps)] |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
402 inputInsertionToN <- s2c(x[2]) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
403 inputInsertionToN[posInsertions]!="-" |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
404 inputInsertionToN[posInsertions] <- "N" |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
405 inputInsertionToN <- c2s(inputInsertionToN) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
406 matInput[rowIndex,2] <<- inputInsertionToN |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
407 }) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
408 return(insPos) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
409 } |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
410 |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
411 # Check sequences for deletions |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
412 checkDeletions <-function(matInputP){ |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
413 deletionCheck = apply( matInputP,1, function(x){ |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
414 inputGaps <- as.vector( gregexpr("-",x[1])[[1]] ) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
415 glGaps <- as.vector( gregexpr("-",x[2])[[1]] ) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
416 return( is.finite( match(FALSE, inputGaps%in%glGaps ) ) ) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
417 }) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
418 return(as.vector(deletionCheck)) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
419 } |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
420 # Fix sequences with deletions |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
421 fixDeletions <- function(matInputP){ |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
422 delPos <- checkDeletions(matInputP) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
423 sapply((1:nrow(matInputP))[delPos],function(rowIndex){ |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
424 x <- matInputP[rowIndex,] |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
425 inputGaps <- gregexpr("-",x[1])[[1]] |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
426 glGaps <- gregexpr("-",x[2])[[1]] |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
427 posDeletions <- inputGaps[!(inputGaps%in%glGaps)] |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
428 inputDeletionToN <- s2c(x[1]) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
429 inputDeletionToN[posDeletions] <- "N" |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
430 inputDeletionToN <- c2s(inputDeletionToN) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
431 matInput[rowIndex,1] <<- inputDeletionToN |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
432 }) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
433 return(delPos) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
434 } |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
435 |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
436 |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
437 # Trim DNA sequence to the last codon |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
438 trimToLastCodon <- function(seqToTrim){ |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
439 seqLen = nchar(seqToTrim) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
440 trimmedSeq = s2c(seqToTrim) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
441 poi = seqLen |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
442 tailLen = 0 |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
443 |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
444 while(trimmedSeq[poi]=="-" || trimmedSeq[poi]=="."){ |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
445 tailLen = tailLen + 1 |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
446 poi = poi - 1 |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
447 } |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
448 |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
449 trimmedSeq = c2s(trimmedSeq[1:(seqLen-tailLen)]) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
450 seqLen = nchar(trimmedSeq) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
451 # Trim sequence to last codon |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
452 if( getCodonPos(seqLen)[3] > seqLen ) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
453 trimmedSeq = substr(seqToTrim,1, ( (getCodonPos(seqLen)[1])-1 ) ) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
454 |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
455 return(trimmedSeq) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
456 } |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
457 |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
458 # Given a nuclotide position, returns the pos of the 3 nucs that made the codon |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
459 # e.g. nuc 86 is part of nucs 85,86,87 |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
460 getCodonPos <- function(nucPos){ |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
461 codonNum = (ceiling(nucPos/3))*3 |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
462 return( (codonNum-2):codonNum) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
463 } |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
464 |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
465 # Given a nuclotide position, returns the codon number |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
466 # e.g. nuc 86 = codon 29 |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
467 getCodonNumb <- function(nucPos){ |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
468 return( ceiling(nucPos/3) ) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
469 } |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
470 |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
471 # Given a codon, returns all the nuc positions that make the codon |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
472 getCodonNucs <- function(codonNumb){ |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
473 getCodonPos(codonNumb*3) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
474 } |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
475 |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
476 computeCodonTable <- function(testID=1){ |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
477 |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
478 if(testID<=4){ |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
479 # Pre-compute every codons |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
480 intCounter = 1 |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
481 for(pOne in NUCLEOTIDES){ |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
482 for(pTwo in NUCLEOTIDES){ |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
483 for(pThree in NUCLEOTIDES){ |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
484 codon = paste(pOne,pTwo,pThree,sep="") |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
485 colnames(CODON_TABLE)[intCounter] = codon |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
486 intCounter = intCounter + 1 |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
487 CODON_TABLE[,codon] = mutationTypeOptimized(cbind(permutateAllCodon(codon),rep(codon,12))) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
488 } |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
489 } |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
490 } |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
491 chars = c("N","A","C","G","T", "-") |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
492 for(a in chars){ |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
493 for(b in chars){ |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
494 for(c in chars){ |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
495 if(a=="N" | b=="N" | c=="N"){ |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
496 #cat(paste(a,b,c),sep="","\n") |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
497 CODON_TABLE[,paste(a,b,c,sep="")] = rep(NA,12) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
498 } |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
499 } |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
500 } |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
501 } |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
502 |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
503 chars = c("-","A","C","G","T") |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
504 for(a in chars){ |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
505 for(b in chars){ |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
506 for(c in chars){ |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
507 if(a=="-" | b=="-" | c=="-"){ |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
508 #cat(paste(a,b,c),sep="","\n") |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
509 CODON_TABLE[,paste(a,b,c,sep="")] = rep(NA,12) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
510 } |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
511 } |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
512 } |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
513 } |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
514 CODON_TABLE <<- as.matrix(CODON_TABLE) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
515 } |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
516 } |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
517 |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
518 collapseClone <- function(vecInputSeqs,glSeq,readEnd,nonTerminalOnly=0){ |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
519 #print(length(vecInputSeqs)) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
520 vecInputSeqs = unique(vecInputSeqs) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
521 if(length(vecInputSeqs)==1){ |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
522 return( list( c(vecInputSeqs,glSeq), F) ) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
523 }else{ |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
524 charInputSeqs <- sapply(vecInputSeqs, function(x){ |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
525 s2c(x)[1:readEnd] |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
526 }) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
527 charGLSeq <- s2c(glSeq) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
528 matClone <- sapply(1:readEnd, function(i){ |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
529 posNucs = unique(charInputSeqs[i,]) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
530 posGL = charGLSeq[i] |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
531 error = FALSE |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
532 if(posGL=="-" & sum(!(posNucs%in%c("-","N")))==0 ){ |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
533 return(c("-",error)) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
534 } |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
535 if(length(posNucs)==1) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
536 return(c(posNucs[1],error)) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
537 else{ |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
538 if("N"%in%posNucs){ |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
539 error=TRUE |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
540 } |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
541 if(sum(!posNucs[posNucs!="N"]%in%posGL)==0){ |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
542 return( c(posGL,error) ) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
543 }else{ |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
544 #return( c(sample(posNucs[posNucs!="N"],1),error) ) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
545 if(nonTerminalOnly==0){ |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
546 return( c(sample(charInputSeqs[i,charInputSeqs[i,]!="N" & charInputSeqs[i,]!=posGL],1),error) ) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
547 }else{ |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
548 posNucs = charInputSeqs[i,charInputSeqs[i,]!="N" & charInputSeqs[i,]!=posGL] |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
549 posNucsTable = table(posNucs) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
550 if(sum(posNucsTable>1)==0){ |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
551 return( c(posGL,error) ) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
552 }else{ |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
553 return( c(sample( posNucs[posNucs%in%names(posNucsTable)[posNucsTable>1]],1),error) ) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
554 } |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
555 } |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
556 |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
557 } |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
558 } |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
559 }) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
560 |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
561 |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
562 #print(length(vecInputSeqs)) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
563 return(list(c(c2s(matClone[1,]),glSeq),"TRUE"%in%matClone[2,])) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
564 } |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
565 } |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
566 |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
567 # Compute the expected for each sequence-germline pair |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
568 getExpectedIndividual <- function(matInput){ |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
569 if( any(grep("multicore",search())) ){ |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
570 facGL <- factor(matInput[,2]) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
571 facLevels = levels(facGL) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
572 LisGLs_MutabilityU = mclapply(1:length(facLevels), function(x){ |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
573 computeMutabilities(facLevels[x]) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
574 }) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
575 facIndex = match(facGL,facLevels) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
576 |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
577 LisGLs_Mutability = mclapply(1:nrow(matInput), function(x){ |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
578 cInput = rep(NA,nchar(matInput[x,1])) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
579 cInput[s2c(matInput[x,1])!="N"] = 1 |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
580 LisGLs_MutabilityU[[facIndex[x]]] * cInput |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
581 }) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
582 |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
583 LisGLs_Targeting = mclapply(1:dim(matInput)[1], function(x){ |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
584 computeTargeting(matInput[x,2],LisGLs_Mutability[[x]]) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
585 }) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
586 |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
587 LisGLs_MutationTypes = mclapply(1:length(matInput[,2]),function(x){ |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
588 #print(x) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
589 computeMutationTypes(matInput[x,2]) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
590 }) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
591 |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
592 LisGLs_Exp = mclapply(1:dim(matInput)[1], function(x){ |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
593 computeExpected(LisGLs_Targeting[[x]],LisGLs_MutationTypes[[x]]) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
594 }) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
595 |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
596 ul_LisGLs_Exp = unlist(LisGLs_Exp) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
597 return(matrix(ul_LisGLs_Exp,ncol=4,nrow=(length(ul_LisGLs_Exp)/4),byrow=T)) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
598 }else{ |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
599 facGL <- factor(matInput[,2]) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
600 facLevels = levels(facGL) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
601 LisGLs_MutabilityU = lapply(1:length(facLevels), function(x){ |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
602 computeMutabilities(facLevels[x]) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
603 }) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
604 facIndex = match(facGL,facLevels) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
605 |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
606 LisGLs_Mutability = lapply(1:nrow(matInput), function(x){ |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
607 cInput = rep(NA,nchar(matInput[x,1])) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
608 cInput[s2c(matInput[x,1])!="N"] = 1 |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
609 LisGLs_MutabilityU[[facIndex[x]]] * cInput |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
610 }) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
611 |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
612 LisGLs_Targeting = lapply(1:dim(matInput)[1], function(x){ |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
613 computeTargeting(matInput[x,2],LisGLs_Mutability[[x]]) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
614 }) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
615 |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
616 LisGLs_MutationTypes = lapply(1:length(matInput[,2]),function(x){ |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
617 #print(x) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
618 computeMutationTypes(matInput[x,2]) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
619 }) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
620 |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
621 LisGLs_Exp = lapply(1:dim(matInput)[1], function(x){ |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
622 computeExpected(LisGLs_Targeting[[x]],LisGLs_MutationTypes[[x]]) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
623 }) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
624 |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
625 ul_LisGLs_Exp = unlist(LisGLs_Exp) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
626 return(matrix(ul_LisGLs_Exp,ncol=4,nrow=(length(ul_LisGLs_Exp)/4),byrow=T)) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
627 |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
628 } |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
629 } |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
630 |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
631 # Compute mutabilities of sequence based on the tri-nucleotide model |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
632 computeMutabilities <- function(paramSeq){ |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
633 seqLen = nchar(paramSeq) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
634 seqMutabilites = rep(NA,seqLen) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
635 |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
636 gaplessSeq = gsub("-", "", paramSeq) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
637 gaplessSeqLen = nchar(gaplessSeq) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
638 gaplessSeqMutabilites = rep(NA,gaplessSeqLen) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
639 |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
640 if(mutabilityModel!=5){ |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
641 pos<- 3:(gaplessSeqLen) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
642 subSeq = substr(rep(gaplessSeq,gaplessSeqLen-2),(pos-2),(pos+2)) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
643 gaplessSeqMutabilites[pos] = |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
644 tapply( c( |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
645 getMutability( substr(subSeq,1,3), 3) , |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
646 getMutability( substr(subSeq,2,4), 2), |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
647 getMutability( substr(subSeq,3,5), 1) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
648 ),rep(1:(gaplessSeqLen-2),3),mean,na.rm=TRUE |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
649 ) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
650 #Pos 1 |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
651 subSeq = substr(gaplessSeq,1,3) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
652 gaplessSeqMutabilites[1] = getMutability(subSeq , 1) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
653 #Pos 2 |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
654 subSeq = substr(gaplessSeq,1,4) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
655 gaplessSeqMutabilites[2] = mean( c( |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
656 getMutability( substr(subSeq,1,3), 2) , |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
657 getMutability( substr(subSeq,2,4), 1) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
658 ),na.rm=T |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
659 ) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
660 seqMutabilites[which(s2c(paramSeq)!="-")]<- gaplessSeqMutabilites |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
661 return(seqMutabilites) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
662 }else{ |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
663 |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
664 pos<- 3:(gaplessSeqLen) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
665 subSeq = substr(rep(gaplessSeq,gaplessSeqLen-2),(pos-2),(pos+2)) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
666 gaplessSeqMutabilites[pos] = sapply(subSeq,function(x){ getMutability5(x) }, simplify=T) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
667 seqMutabilites[which(s2c(paramSeq)!="-")]<- gaplessSeqMutabilites |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
668 return(seqMutabilites) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
669 } |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
670 |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
671 } |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
672 |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
673 # Returns the mutability of a triplet at a given position |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
674 getMutability <- function(codon, pos=1:3){ |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
675 triplets <- rownames(mutability) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
676 mutability[ match(codon,triplets) ,pos] |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
677 } |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
678 |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
679 getMutability5 <- function(fivemer){ |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
680 return(mutability[fivemer]) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
681 } |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
682 |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
683 # Returns the substitution probabilty |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
684 getTransistionProb <- function(nuc){ |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
685 substitution[nuc,] |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
686 } |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
687 |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
688 getTransistionProb5 <- function(fivemer){ |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
689 if(any(which(fivemer==colnames(substitution)))){ |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
690 return(substitution[,fivemer]) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
691 }else{ |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
692 return(array(NA,4)) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
693 } |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
694 } |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
695 |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
696 # Given a nuc, returns the other 3 nucs it can mutate to |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
697 canMutateTo <- function(nuc){ |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
698 NUCLEOTIDES[- which(NUCLEOTIDES==nuc)] |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
699 } |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
700 |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
701 # Given a nucleotide, returns the probabilty of other nucleotide it can mutate to |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
702 canMutateToProb <- function(nuc){ |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
703 substitution[nuc,canMutateTo(nuc)] |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
704 } |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
705 |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
706 # Compute targeting, based on precomputed mutatbility & substitution |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
707 computeTargeting <- function(param_strSeq,param_vecMutabilities){ |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
708 |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
709 if(substitutionModel!=5){ |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
710 vecSeq = s2c(param_strSeq) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
711 matTargeting = sapply( 1:length(vecSeq), function(x) { param_vecMutabilities[x] * getTransistionProb(vecSeq[x]) } ) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
712 #matTargeting = apply( rbind(vecSeq,param_vecMutabilities),2, function(x) { as.vector(as.numeric(x[2]) * getTransistionProb(x[1])) } ) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
713 dimnames( matTargeting ) = list(NUCLEOTIDES,1:(length(vecSeq))) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
714 return (matTargeting) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
715 }else{ |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
716 |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
717 seqLen = nchar(param_strSeq) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
718 seqsubstitution = matrix(NA,ncol=seqLen,nrow=4) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
719 paramSeq <- param_strSeq |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
720 gaplessSeq = gsub("-", "", paramSeq) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
721 gaplessSeqLen = nchar(gaplessSeq) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
722 gaplessSeqSubstitution = matrix(NA,ncol=gaplessSeqLen,nrow=4) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
723 |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
724 pos<- 3:(gaplessSeqLen) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
725 subSeq = substr(rep(gaplessSeq,gaplessSeqLen-2),(pos-2),(pos+2)) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
726 gaplessSeqSubstitution[,pos] = sapply(subSeq,function(x){ getTransistionProb5(x) }, simplify=T) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
727 seqsubstitution[,which(s2c(paramSeq)!="-")]<- gaplessSeqSubstitution |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
728 #matTargeting <- param_vecMutabilities %*% seqsubstitution |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
729 matTargeting <- sweep(seqsubstitution,2,param_vecMutabilities,`*`) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
730 dimnames( matTargeting ) = list(NUCLEOTIDES,1:(seqLen)) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
731 return (matTargeting) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
732 } |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
733 } |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
734 |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
735 # Compute the mutations types |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
736 computeMutationTypes <- function(param_strSeq){ |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
737 #cat(param_strSeq,"\n") |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
738 #vecSeq = trimToLastCodon(param_strSeq) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
739 lenSeq = nchar(param_strSeq) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
740 vecCodons = sapply({1:(lenSeq/3)}*3-2,function(x){substr(param_strSeq,x,x+2)}) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
741 matMutationTypes = matrix( unlist(CODON_TABLE[,vecCodons]) ,ncol=lenSeq,nrow=4, byrow=F) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
742 dimnames( matMutationTypes ) = list(NUCLEOTIDES,1:(ncol(matMutationTypes))) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
743 return(matMutationTypes) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
744 } |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
745 computeMutationTypesFast <- function(param_strSeq){ |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
746 matMutationTypes = matrix( CODON_TABLE[,param_strSeq] ,ncol=3,nrow=4, byrow=F) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
747 #dimnames( matMutationTypes ) = list(NUCLEOTIDES,1:(length(vecSeq))) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
748 return(matMutationTypes) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
749 } |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
750 mutationTypeOptimized <- function( matOfCodons ){ |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
751 apply( matOfCodons,1,function(x){ mutationType(x[2],x[1]) } ) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
752 } |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
753 |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
754 # Returns a vector of codons 1 mutation away from the given codon |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
755 permutateAllCodon <- function(codon){ |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
756 cCodon = s2c(codon) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
757 matCodons = t(array(cCodon,dim=c(3,12))) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
758 matCodons[1:4,1] = NUCLEOTIDES |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
759 matCodons[5:8,2] = NUCLEOTIDES |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
760 matCodons[9:12,3] = NUCLEOTIDES |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
761 apply(matCodons,1,c2s) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
762 } |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
763 |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
764 # Given two codons, tells you if the mutation is R or S (based on your definition) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
765 mutationType <- function(codonFrom,codonTo){ |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
766 if(testID==4){ |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
767 if( is.na(codonFrom) | is.na(codonTo) | is.na(translateCodonToAminoAcid(codonFrom)) | is.na(translateCodonToAminoAcid(codonTo)) ){ |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
768 return(NA) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
769 }else{ |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
770 mutationType = "S" |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
771 if( translateAminoAcidToTraitChange(translateCodonToAminoAcid(codonFrom)) != translateAminoAcidToTraitChange(translateCodonToAminoAcid(codonTo)) ){ |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
772 mutationType = "R" |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
773 } |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
774 if(translateCodonToAminoAcid(codonTo)=="*" | translateCodonToAminoAcid(codonFrom)=="*"){ |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
775 mutationType = "Stop" |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
776 } |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
777 return(mutationType) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
778 } |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
779 }else if(testID==5){ |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
780 if( is.na(codonFrom) | is.na(codonTo) | is.na(translateCodonToAminoAcid(codonFrom)) | is.na(translateCodonToAminoAcid(codonTo)) ){ |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
781 return(NA) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
782 }else{ |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
783 if(codonFrom==codonTo){ |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
784 mutationType = "S" |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
785 }else{ |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
786 codonFrom = s2c(codonFrom) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
787 codonTo = s2c(codonTo) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
788 mutationType = "Stop" |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
789 nucOfI = codonFrom[which(codonTo!=codonFrom)] |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
790 if(nucOfI=="C"){ |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
791 mutationType = "R" |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
792 }else if(nucOfI=="G"){ |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
793 mutationType = "S" |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
794 } |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
795 } |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
796 return(mutationType) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
797 } |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
798 }else{ |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
799 if( is.na(codonFrom) | is.na(codonTo) | is.na(translateCodonToAminoAcid(codonFrom)) | is.na(translateCodonToAminoAcid(codonTo)) ){ |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
800 return(NA) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
801 }else{ |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
802 mutationType = "S" |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
803 if( translateCodonToAminoAcid(codonFrom) != translateCodonToAminoAcid(codonTo) ){ |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
804 mutationType = "R" |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
805 } |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
806 if(translateCodonToAminoAcid(codonTo)=="*" | translateCodonToAminoAcid(codonFrom)=="*"){ |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
807 mutationType = "Stop" |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
808 } |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
809 return(mutationType) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
810 } |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
811 } |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
812 } |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
813 |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
814 |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
815 #given a mat of targeting & it's corresponding mutationtypes returns |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
816 #a vector of Exp_RCDR,Exp_SCDR,Exp_RFWR,Exp_RFWR |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
817 computeExpected <- function(paramTargeting,paramMutationTypes){ |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
818 # Replacements |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
819 RPos = which(paramMutationTypes=="R") |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
820 #FWR |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
821 Exp_R_FWR = sum(paramTargeting[ RPos[which(FWR_Nuc_Mat[RPos]==T)] ],na.rm=T) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
822 #CDR |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
823 Exp_R_CDR = sum(paramTargeting[ RPos[which(CDR_Nuc_Mat[RPos]==T)] ],na.rm=T) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
824 # Silents |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
825 SPos = which(paramMutationTypes=="S") |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
826 #FWR |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
827 Exp_S_FWR = sum(paramTargeting[ SPos[which(FWR_Nuc_Mat[SPos]==T)] ],na.rm=T) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
828 #CDR |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
829 Exp_S_CDR = sum(paramTargeting[ SPos[which(CDR_Nuc_Mat[SPos]==T)] ],na.rm=T) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
830 |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
831 return(c(Exp_R_CDR,Exp_S_CDR,Exp_R_FWR,Exp_S_FWR)) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
832 } |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
833 |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
834 # Count the mutations in a sequence |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
835 # each mutation is treated independently |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
836 analyzeMutations2NucUri_website <- function( rev_in_matrix ){ |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
837 paramGL = rev_in_matrix[2,] |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
838 paramSeq = rev_in_matrix[1,] |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
839 |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
840 #Fill seq with GL seq if gapped |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
841 #if( any(paramSeq=="-") ){ |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
842 # gapPos_Seq = which(paramSeq=="-") |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
843 # gapPos_Seq_ToReplace = gapPos_Seq[paramGL[gapPos_Seq] != "-"] |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
844 # paramSeq[gapPos_Seq_ToReplace] = paramGL[gapPos_Seq_ToReplace] |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
845 #} |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
846 |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
847 |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
848 #if( any(paramSeq=="N") ){ |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
849 # gapPos_Seq = which(paramSeq=="N") |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
850 # gapPos_Seq_ToReplace = gapPos_Seq[paramGL[gapPos_Seq] != "N"] |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
851 # paramSeq[gapPos_Seq_ToReplace] = paramGL[gapPos_Seq_ToReplace] |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
852 #} |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
853 |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
854 analyzeMutations2NucUri( matrix(c( paramGL, paramSeq ),2,length(paramGL),byrow=T) ) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
855 |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
856 } |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
857 |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
858 #1 = GL |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
859 #2 = Seq |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
860 analyzeMutations2NucUri <- function( in_matrix=matrix(c(c("A","A","A","C","C","C"),c("A","G","G","C","C","A")),2,6,byrow=T) ){ |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
861 paramGL = in_matrix[2,] |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
862 paramSeq = in_matrix[1,] |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
863 paramSeqUri = paramGL |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
864 #mutations = apply(rbind(paramGL,paramSeq), 2, function(x){!x[1]==x[2]}) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
865 mutations_val = paramGL != paramSeq |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
866 if(any(mutations_val)){ |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
867 mutationPos = {1:length(mutations_val)}[mutations_val] |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
868 mutationPos = mutationPos[sapply(mutationPos, function(x){!any(paramSeq[getCodonPos(x)]=="N")})] |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
869 length_mutations =length(mutationPos) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
870 mutationInfo = rep(NA,length_mutations) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
871 if(any(mutationPos)){ |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
872 |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
873 pos<- mutationPos |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
874 pos_array<-array(sapply(pos,getCodonPos)) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
875 codonGL = paramGL[pos_array] |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
876 |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
877 codonSeq = sapply(pos,function(x){ |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
878 seqP = paramGL[getCodonPos(x)] |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
879 muCodonPos = {x-1}%%3+1 |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
880 seqP[muCodonPos] = paramSeq[x] |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
881 return(seqP) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
882 }) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
883 GLcodons = apply(matrix(codonGL,length_mutations,3,byrow=TRUE),1,c2s) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
884 Seqcodons = apply(codonSeq,2,c2s) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
885 mutationInfo = apply(rbind(GLcodons , Seqcodons),2,function(x){mutationType(c2s(x[1]),c2s(x[2]))}) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
886 names(mutationInfo) = mutationPos |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
887 } |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
888 if(any(!is.na(mutationInfo))){ |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
889 return(mutationInfo[!is.na(mutationInfo)]) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
890 }else{ |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
891 return(NA) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
892 } |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
893 |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
894 |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
895 }else{ |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
896 return (NA) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
897 } |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
898 } |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
899 |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
900 processNucMutations2 <- function(mu){ |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
901 if(!is.na(mu)){ |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
902 #R |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
903 if(any(mu=="R")){ |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
904 Rs = mu[mu=="R"] |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
905 nucNumbs = as.numeric(names(Rs)) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
906 R_CDR = sum(as.integer(CDR_Nuc[nucNumbs]),na.rm=T) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
907 R_FWR = sum(as.integer(FWR_Nuc[nucNumbs]),na.rm=T) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
908 }else{ |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
909 R_CDR = 0 |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
910 R_FWR = 0 |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
911 } |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
912 |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
913 #S |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
914 if(any(mu=="S")){ |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
915 Ss = mu[mu=="S"] |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
916 nucNumbs = as.numeric(names(Ss)) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
917 S_CDR = sum(as.integer(CDR_Nuc[nucNumbs]),na.rm=T) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
918 S_FWR = sum(as.integer(FWR_Nuc[nucNumbs]),na.rm=T) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
919 }else{ |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
920 S_CDR = 0 |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
921 S_FWR = 0 |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
922 } |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
923 |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
924 |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
925 retVec = c(R_CDR,S_CDR,R_FWR,S_FWR) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
926 retVec[is.na(retVec)]=0 |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
927 return(retVec) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
928 }else{ |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
929 return(rep(0,4)) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
930 } |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
931 } |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
932 |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
933 |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
934 ## Z-score Test |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
935 computeZScore <- function(mat, test="Focused"){ |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
936 matRes <- matrix(NA,ncol=2,nrow=(nrow(mat))) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
937 if(test=="Focused"){ |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
938 #Z_Focused_CDR |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
939 #P_Denom = sum( mat[1,c(5,6,8)], na.rm=T ) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
940 P = apply(mat[,c(5,6,8)],1,function(x){(x[1]/sum(x))}) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
941 R_mean = apply(cbind(mat[,c(1,2,4)],P),1,function(x){x[4]*(sum(x[1:3]))}) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
942 R_sd=sqrt(R_mean*(1-P)) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
943 matRes[,1] = (mat[,1]-R_mean)/R_sd |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
944 |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
945 #Z_Focused_FWR |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
946 #P_Denom = sum( mat[1,c(7,6,8)], na.rm=T ) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
947 P = apply(mat[,c(7,6,8)],1,function(x){(x[1]/sum(x))}) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
948 R_mean = apply(cbind(mat[,c(3,2,4)],P),1,function(x){x[4]*(sum(x[1:3]))}) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
949 R_sd=sqrt(R_mean*(1-P)) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
950 matRes[,2] = (mat[,3]-R_mean)/R_sd |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
951 } |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
952 |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
953 if(test=="Local"){ |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
954 #Z_Focused_CDR |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
955 #P_Denom = sum( mat[1,c(5,6,8)], na.rm=T ) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
956 P = apply(mat[,c(5,6)],1,function(x){(x[1]/sum(x))}) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
957 R_mean = apply(cbind(mat[,c(1,2)],P),1,function(x){x[3]*(sum(x[1:2]))}) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
958 R_sd=sqrt(R_mean*(1-P)) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
959 matRes[,1] = (mat[,1]-R_mean)/R_sd |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
960 |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
961 #Z_Focused_FWR |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
962 #P_Denom = sum( mat[1,c(7,6,8)], na.rm=T ) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
963 P = apply(mat[,c(7,8)],1,function(x){(x[1]/sum(x))}) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
964 R_mean = apply(cbind(mat[,c(3,4)],P),1,function(x){x[3]*(sum(x[1:2]))}) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
965 R_sd=sqrt(R_mean*(1-P)) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
966 matRes[,2] = (mat[,3]-R_mean)/R_sd |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
967 } |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
968 |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
969 if(test=="Imbalanced"){ |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
970 #Z_Focused_CDR |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
971 #P_Denom = sum( mat[1,c(5,6,8)], na.rm=T ) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
972 P = apply(mat[,5:8],1,function(x){((x[1]+x[2])/sum(x))}) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
973 R_mean = apply(cbind(mat[,1:4],P),1,function(x){x[5]*(sum(x[1:4]))}) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
974 R_sd=sqrt(R_mean*(1-P)) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
975 matRes[,1] = (mat[,1]-R_mean)/R_sd |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
976 |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
977 #Z_Focused_FWR |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
978 #P_Denom = sum( mat[1,c(7,6,8)], na.rm=T ) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
979 P = apply(mat[,5:8],1,function(x){((x[3]+x[4])/sum(x))}) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
980 R_mean = apply(cbind(mat[,1:4],P),1,function(x){x[5]*(sum(x[1:4]))}) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
981 R_sd=sqrt(R_mean*(1-P)) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
982 matRes[,2] = (mat[,3]-R_mean)/R_sd |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
983 } |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
984 |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
985 matRes[is.nan(matRes)] = NA |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
986 return(matRes) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
987 } |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
988 |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
989 # Return a p-value for a z-score |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
990 z2p <- function(z){ |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
991 p=NA |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
992 if( !is.nan(z) && !is.na(z)){ |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
993 if(z>0){ |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
994 p = (1 - pnorm(z,0,1)) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
995 } else if(z<0){ |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
996 p = (-1 * pnorm(z,0,1)) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
997 } else{ |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
998 p = 0.5 |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
999 } |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1000 }else{ |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1001 p = NA |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1002 } |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1003 return(p) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1004 } |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1005 |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1006 |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1007 ## Bayesian Test |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1008 |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1009 # Fitted parameter for the bayesian framework |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1010 BAYESIAN_FITTED<-c(0.407277142798302, 0.554007336744485, 0.63777155771234, 0.693989162719009, 0.735450014674917, 0.767972534429806, 0.794557287143399, 0.816906816601605, 0.83606796225341, 0.852729446430296, 0.867370424541641, 0.880339760590323, 0.891900995024999, 0.902259181289864, 0.911577919359,0.919990301665853, 0.927606458124537, 0.934518806350661, 0.940805863754375, 0.946534836475715, 0.951763691199255, 0.95654428191308, 0.960920179487397, 0.964930893680829, 0.968611312149038, 0.971992459313836, 0.975102110004818, 0.977964943023096, 0.980603428208439, 0.983037660179428, 0.985285800977406, 0.987364285326685, 0.989288037855441, 0.991070478823525, 0.992723699729969, 0.994259575477392, 0.995687688867975, 0.997017365051493, 0.998257085153047, 0.999414558305388, 1.00049681357804, 1.00151036237481, 1.00246080204981, 1.00335370751909, 1.0041939329768, 1.0049859393417, 1.00573382091263, 1.00644127217376, 1.00711179729107, 1.00774845526417, 1.00835412715854, 1.00893143010366, 1.00948275846309, 1.01001030293661, 1.01051606798079, 1.01100188771288, 1.01146944044216, 1.01192026195449, 1.01235575766094, 1.01277721370986) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1011 CONST_i <- sort(c(((2^(seq(-39,0,length.out=201)))/2)[1:200],(c(0:11,13:99)+0.5)/100,1-(2^(seq(-39,0,length.out=201)))/2)) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1012 |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1013 # Given x, M & p, returns a pdf |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1014 calculate_bayes <- function ( x=3, N=10, p=0.33, |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1015 i=CONST_i, |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1016 max_sigma=20,length_sigma=4001 |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1017 ){ |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1018 if(!0%in%N){ |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1019 G <- max(length(x),length(N),length(p)) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1020 x=array(x,dim=G) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1021 N=array(N,dim=G) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1022 p=array(p,dim=G) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1023 sigma_s<-seq(-max_sigma,max_sigma,length.out=length_sigma) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1024 sigma_1<-log({i/{1-i}}/{p/{1-p}}) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1025 index<-min(N,60) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1026 y<-dbeta(i,x+BAYESIAN_FITTED[index],N+BAYESIAN_FITTED[index]-x)*(1-p)*p*exp(sigma_1)/({1-p}^2+2*p*{1-p}*exp(sigma_1)+{p^2}*exp(2*sigma_1)) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1027 if(!sum(is.na(y))){ |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1028 tmp<-approx(sigma_1,y,sigma_s)$y |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1029 tmp/sum(tmp)/{2*max_sigma/{length_sigma-1}} |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1030 }else{ |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1031 return(NA) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1032 } |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1033 }else{ |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1034 return(NA) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1035 } |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1036 } |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1037 # Given a mat of observed & expected, return a list of CDR & FWR pdf for selection |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1038 computeBayesianScore <- function(mat, test="Focused", max_sigma=20,length_sigma=4001){ |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1039 flagOneSeq = F |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1040 if(nrow(mat)==1){ |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1041 mat=rbind(mat,mat) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1042 flagOneSeq = T |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1043 } |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1044 if(test=="Focused"){ |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1045 #CDR |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1046 P = c(apply(mat[,c(5,6,8)],1,function(x){(x[1]/sum(x))}),0.5) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1047 N = c(apply(mat[,c(1,2,4)],1,function(x){(sum(x))}),0) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1048 X = c(mat[,1],0) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1049 bayesCDR = apply(cbind(X,N,P),1,function(x){calculate_bayes(x=x[1],N=x[2],p=x[3],max_sigma=max_sigma,length_sigma=length_sigma)}) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1050 bayesCDR = bayesCDR[-length(bayesCDR)] |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1051 |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1052 #FWR |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1053 P = c(apply(mat[,c(7,6,8)],1,function(x){(x[1]/sum(x))}),0.5) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1054 N = c(apply(mat[,c(3,2,4)],1,function(x){(sum(x))}),0) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1055 X = c(mat[,3],0) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1056 bayesFWR = apply(cbind(X,N,P),1,function(x){calculate_bayes(x=x[1],N=x[2],p=x[3],max_sigma=max_sigma,length_sigma=length_sigma)}) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1057 bayesFWR = bayesFWR[-length(bayesFWR)] |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1058 } |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1059 |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1060 if(test=="Local"){ |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1061 #CDR |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1062 P = c(apply(mat[,c(5,6)],1,function(x){(x[1]/sum(x))}),0.5) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1063 N = c(apply(mat[,c(1,2)],1,function(x){(sum(x))}),0) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1064 X = c(mat[,1],0) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1065 bayesCDR = apply(cbind(X,N,P),1,function(x){calculate_bayes(x=x[1],N=x[2],p=x[3],max_sigma=max_sigma,length_sigma=length_sigma)}) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1066 bayesCDR = bayesCDR[-length(bayesCDR)] |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1067 |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1068 #FWR |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1069 P = c(apply(mat[,c(7,8)],1,function(x){(x[1]/sum(x))}),0.5) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1070 N = c(apply(mat[,c(3,4)],1,function(x){(sum(x))}),0) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1071 X = c(mat[,3],0) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1072 bayesFWR = apply(cbind(X,N,P),1,function(x){calculate_bayes(x=x[1],N=x[2],p=x[3],max_sigma=max_sigma,length_sigma=length_sigma)}) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1073 bayesFWR = bayesFWR[-length(bayesFWR)] |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1074 } |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1075 |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1076 if(test=="Imbalanced"){ |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1077 #CDR |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1078 P = c(apply(mat[,c(5:8)],1,function(x){((x[1]+x[2])/sum(x))}),0.5) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1079 N = c(apply(mat[,c(1:4)],1,function(x){(sum(x))}),0) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1080 X = c(apply(mat[,c(1:2)],1,function(x){(sum(x))}),0) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1081 bayesCDR = apply(cbind(X,N,P),1,function(x){calculate_bayes(x=x[1],N=x[2],p=x[3],max_sigma=max_sigma,length_sigma=length_sigma)}) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1082 bayesCDR = bayesCDR[-length(bayesCDR)] |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1083 |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1084 #FWR |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1085 P = c(apply(mat[,c(5:8)],1,function(x){((x[3]+x[4])/sum(x))}),0.5) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1086 N = c(apply(mat[,c(1:4)],1,function(x){(sum(x))}),0) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1087 X = c(apply(mat[,c(3:4)],1,function(x){(sum(x))}),0) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1088 bayesFWR = apply(cbind(X,N,P),1,function(x){calculate_bayes(x=x[1],N=x[2],p=x[3],max_sigma=max_sigma,length_sigma=length_sigma)}) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1089 bayesFWR = bayesFWR[-length(bayesFWR)] |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1090 } |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1091 |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1092 if(test=="ImbalancedSilent"){ |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1093 #CDR |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1094 P = c(apply(mat[,c(6,8)],1,function(x){((x[1])/sum(x))}),0.5) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1095 N = c(apply(mat[,c(2,4)],1,function(x){(sum(x))}),0) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1096 X = c(apply(mat[,c(2,4)],1,function(x){(x[1])}),0) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1097 bayesCDR = apply(cbind(X,N,P),1,function(x){calculate_bayes(x=x[1],N=x[2],p=x[3],max_sigma=max_sigma,length_sigma=length_sigma)}) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1098 bayesCDR = bayesCDR[-length(bayesCDR)] |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1099 |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1100 #FWR |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1101 P = c(apply(mat[,c(6,8)],1,function(x){((x[2])/sum(x))}),0.5) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1102 N = c(apply(mat[,c(2,4)],1,function(x){(sum(x))}),0) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1103 X = c(apply(mat[,c(2,4)],1,function(x){(x[2])}),0) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1104 bayesFWR = apply(cbind(X,N,P),1,function(x){calculate_bayes(x=x[1],N=x[2],p=x[3],max_sigma=max_sigma,length_sigma=length_sigma)}) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1105 bayesFWR = bayesFWR[-length(bayesFWR)] |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1106 } |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1107 |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1108 if(flagOneSeq==T){ |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1109 bayesCDR = bayesCDR[1] |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1110 bayesFWR = bayesFWR[1] |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1111 } |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1112 return( list("CDR"=bayesCDR, "FWR"=bayesFWR) ) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1113 } |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1114 |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1115 ##Covolution |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1116 break2chunks<-function(G=1000){ |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1117 base<-2^round(log(sqrt(G),2),0) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1118 return(c(rep(base,floor(G/base)-1),base+G-(floor(G/base)*base))) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1119 } |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1120 |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1121 PowersOfTwo <- function(G=100){ |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1122 exponents <- array() |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1123 i = 0 |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1124 while(G > 0){ |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1125 i=i+1 |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1126 exponents[i] <- floor( log2(G) ) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1127 G <- G-2^exponents[i] |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1128 } |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1129 return(exponents) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1130 } |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1131 |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1132 convolutionPowersOfTwo <- function( cons, length_sigma=4001 ){ |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1133 G = ncol(cons) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1134 if(G>1){ |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1135 for(gen in log(G,2):1){ |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1136 ll<-seq(from=2,to=2^gen,by=2) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1137 sapply(ll,function(l){cons[,l/2]<<-weighted_conv(cons[,l],cons[,l-1],length_sigma=length_sigma)}) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1138 } |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1139 } |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1140 return( cons[,1] ) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1141 } |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1142 |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1143 convolutionPowersOfTwoByTwos <- function( cons, length_sigma=4001,G=1 ){ |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1144 if(length(ncol(cons))) G<-ncol(cons) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1145 groups <- PowersOfTwo(G) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1146 matG <- matrix(NA, ncol=length(groups), nrow=length(cons)/G ) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1147 startIndex = 1 |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1148 for( i in 1:length(groups) ){ |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1149 stopIndex <- 2^groups[i] + startIndex - 1 |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1150 if(stopIndex!=startIndex){ |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1151 matG[,i] <- convolutionPowersOfTwo( cons[,startIndex:stopIndex], length_sigma=length_sigma ) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1152 startIndex = stopIndex + 1 |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1153 } |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1154 else { |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1155 if(G>1) matG[,i] <- cons[,startIndex:stopIndex] |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1156 else matG[,i] <- cons |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1157 #startIndex = stopIndex + 1 |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1158 } |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1159 } |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1160 return( list( matG, groups ) ) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1161 } |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1162 |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1163 weighted_conv<-function(x,y,w=1,m=100,length_sigma=4001){ |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1164 lx<-length(x) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1165 ly<-length(y) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1166 if({lx<m}| {{lx*w}<m}| {{ly}<m}| {{ly*w}<m}){ |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1167 if(w<1){ |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1168 y1<-approx(1:ly,y,seq(1,ly,length.out=m))$y |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1169 x1<-approx(1:lx,x,seq(1,lx,length.out=m/w))$y |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1170 lx<-length(x1) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1171 ly<-length(y1) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1172 } |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1173 else { |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1174 y1<-approx(1:ly,y,seq(1,ly,length.out=m*w))$y |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1175 x1<-approx(1:lx,x,seq(1,lx,length.out=m))$y |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1176 lx<-length(x1) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1177 ly<-length(y1) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1178 } |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1179 } |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1180 else{ |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1181 x1<-x |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1182 y1<-approx(1:ly,y,seq(1,ly,length.out=floor(lx*w)))$y |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1183 ly<-length(y1) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1184 } |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1185 tmp<-approx(x=1:(lx+ly-1),y=convolve(x1,rev(y1),type="open"),xout=seq(1,lx+ly-1,length.out=length_sigma))$y |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1186 tmp[tmp<=0] = 0 |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1187 return(tmp/sum(tmp)) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1188 } |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1189 |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1190 calculate_bayesGHelper <- function( listMatG,length_sigma=4001 ){ |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1191 matG <- listMatG[[1]] |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1192 groups <- listMatG[[2]] |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1193 i = 1 |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1194 resConv <- matG[,i] |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1195 denom <- 2^groups[i] |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1196 if(length(groups)>1){ |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1197 while( i<length(groups) ){ |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1198 i = i + 1 |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1199 resConv <- weighted_conv(resConv, matG[,i], w= {{2^groups[i]}/denom} ,length_sigma=length_sigma) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1200 #cat({{2^groups[i]}/denom},"\n") |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1201 denom <- denom + 2^groups[i] |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1202 } |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1203 } |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1204 return(resConv) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1205 } |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1206 |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1207 # Given a list of PDFs, returns a convoluted PDF |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1208 groupPosteriors <- function( listPosteriors, max_sigma=20, length_sigma=4001 ,Threshold=2 ){ |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1209 listPosteriors = listPosteriors[ !is.na(listPosteriors) ] |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1210 Length_Postrior<-length(listPosteriors) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1211 if(Length_Postrior>1 & Length_Postrior<=Threshold){ |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1212 cons = matrix(unlist(listPosteriors),length(listPosteriors[[1]]),length(listPosteriors)) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1213 listMatG <- convolutionPowersOfTwoByTwos(cons,length_sigma=length_sigma) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1214 y<-calculate_bayesGHelper(listMatG,length_sigma=length_sigma) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1215 return( y/sum(y)/(2*max_sigma/(length_sigma-1)) ) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1216 }else if(Length_Postrior==1) return(listPosteriors[[1]]) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1217 else if(Length_Postrior==0) return(NA) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1218 else { |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1219 cons = matrix(unlist(listPosteriors),length(listPosteriors[[1]]),length(listPosteriors)) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1220 y = fastConv(cons,max_sigma=max_sigma, length_sigma=length_sigma ) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1221 return( y/sum(y)/(2*max_sigma/(length_sigma-1)) ) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1222 } |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1223 } |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1224 |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1225 fastConv<-function(cons, max_sigma=20, length_sigma=4001){ |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1226 chunks<-break2chunks(G=ncol(cons)) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1227 if(ncol(cons)==3) chunks<-2:1 |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1228 index_chunks_end <- cumsum(chunks) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1229 index_chunks_start <- c(1,index_chunks_end[-length(index_chunks_end)]+1) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1230 index_chunks <- cbind(index_chunks_start,index_chunks_end) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1231 |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1232 case <- sum(chunks!=chunks[1]) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1233 if(case==1) End <- max(1,((length(index_chunks)/2)-1)) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1234 else End <- max(1,((length(index_chunks)/2))) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1235 |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1236 firsts <- sapply(1:End,function(i){ |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1237 indexes<-index_chunks[i,1]:index_chunks[i,2] |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1238 convolutionPowersOfTwoByTwos(cons[ ,indexes])[[1]] |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1239 }) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1240 if(case==0){ |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1241 result<-calculate_bayesGHelper( convolutionPowersOfTwoByTwos(firsts) ) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1242 }else if(case==1){ |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1243 last<-list(calculate_bayesGHelper( |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1244 convolutionPowersOfTwoByTwos( cons[ ,index_chunks[length(index_chunks)/2,1]:index_chunks[length(index_chunks)/2,2]] ) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1245 ),0) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1246 result_first<-calculate_bayesGHelper(convolutionPowersOfTwoByTwos(firsts)) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1247 result<-calculate_bayesGHelper( |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1248 list( |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1249 cbind( |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1250 result_first,last[[1]]), |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1251 c(log(index_chunks_end[length(index_chunks)/2-1],2),log(index_chunks[length(index_chunks)/2,2]-index_chunks[length(index_chunks)/2,1]+1,2)) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1252 ) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1253 ) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1254 } |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1255 return(as.vector(result)) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1256 } |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1257 |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1258 # Computes the 95% CI for a pdf |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1259 calcBayesCI <- function(Pdf,low=0.025,up=0.975,max_sigma=20, length_sigma=4001){ |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1260 if(length(Pdf)!=length_sigma) return(NA) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1261 sigma_s=seq(-max_sigma,max_sigma,length.out=length_sigma) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1262 cdf = cumsum(Pdf) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1263 cdf = cdf/cdf[length(cdf)] |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1264 return( c(sigma_s[findInterval(low,cdf)-1] , sigma_s[findInterval(up,cdf)]) ) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1265 } |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1266 |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1267 # Computes a mean for a pdf |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1268 calcBayesMean <- function(Pdf,max_sigma=20,length_sigma=4001){ |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1269 if(length(Pdf)!=length_sigma) return(NA) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1270 sigma_s=seq(-max_sigma,max_sigma,length.out=length_sigma) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1271 norm = {length_sigma-1}/2/max_sigma |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1272 return( (Pdf%*%sigma_s/norm) ) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1273 } |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1274 |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1275 # Returns the mean, and the 95% CI for a pdf |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1276 calcBayesOutputInfo <- function(Pdf,low=0.025,up=0.975,max_sigma=20, length_sigma=4001){ |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1277 if(is.na(Pdf)) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1278 return(rep(NA,3)) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1279 bCI = calcBayesCI(Pdf=Pdf,low=low,up=up,max_sigma=max_sigma,length_sigma=length_sigma) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1280 bMean = calcBayesMean(Pdf=Pdf,max_sigma=max_sigma,length_sigma=length_sigma) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1281 return(c(bMean, bCI)) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1282 } |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1283 |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1284 # Computes the p-value of a pdf |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1285 computeSigmaP <- function(Pdf, length_sigma=4001, max_sigma=20){ |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1286 if(length(Pdf)>1){ |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1287 norm = {length_sigma-1}/2/max_sigma |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1288 pVal = {sum(Pdf[1:{{length_sigma-1}/2}]) + Pdf[{{length_sigma+1}/2}]/2}/norm |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1289 if(pVal>0.5){ |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1290 pVal = pVal-1 |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1291 } |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1292 return(pVal) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1293 }else{ |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1294 return(NA) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1295 } |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1296 } |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1297 |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1298 # Compute p-value of two distributions |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1299 compareTwoDistsFaster <-function(sigma_S=seq(-20,20,length.out=4001), N=10000, dens1=runif(4001,0,1), dens2=runif(4001,0,1)){ |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1300 #print(c(length(dens1),length(dens2))) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1301 if(length(dens1)>1 & length(dens2)>1 ){ |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1302 dens1<-dens1/sum(dens1) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1303 dens2<-dens2/sum(dens2) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1304 cum2 <- cumsum(dens2)-dens2/2 |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1305 tmp<- sum(sapply(1:length(dens1),function(i)return(dens1[i]*cum2[i]))) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1306 #print(tmp) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1307 if(tmp>0.5)tmp<-tmp-1 |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1308 return( tmp ) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1309 } |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1310 else { |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1311 return(NA) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1312 } |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1313 #return (sum(sapply(1:N,function(i)(sample(sigma_S,1,prob=dens1)>sample(sigma_S,1,prob=dens2))))/N) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1314 } |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1315 |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1316 # get number of seqeunces contributing to the sigma (i.e. seqeunces with mutations) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1317 numberOfSeqsWithMutations <- function(matMutations,test=1){ |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1318 if(test==4)test=2 |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1319 cdrSeqs <- 0 |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1320 fwrSeqs <- 0 |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1321 if(test==1){#focused |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1322 cdrMutations <- apply(matMutations, 1, function(x){ sum(x[c(1,2,4)]) }) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1323 fwrMutations <- apply(matMutations, 1, function(x){ sum(x[c(3,4,2)]) }) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1324 if( any(which(cdrMutations>0)) ) cdrSeqs <- sum(cdrMutations>0) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1325 if( any(which(fwrMutations>0)) ) fwrSeqs <- sum(fwrMutations>0) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1326 } |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1327 if(test==2){#local |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1328 cdrMutations <- apply(matMutations, 1, function(x){ sum(x[c(1,2)]) }) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1329 fwrMutations <- apply(matMutations, 1, function(x){ sum(x[c(3,4)]) }) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1330 if( any(which(cdrMutations>0)) ) cdrSeqs <- sum(cdrMutations>0) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1331 if( any(which(fwrMutations>0)) ) fwrSeqs <- sum(fwrMutations>0) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1332 } |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1333 return(c("CDR"=cdrSeqs, "FWR"=fwrSeqs)) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1334 } |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1335 |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1336 |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1337 |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1338 shadeColor <- function(sigmaVal=NA,pVal=NA){ |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1339 if(is.na(sigmaVal) & is.na(pVal)) return(NA) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1340 if(is.na(sigmaVal) & !is.na(pVal)) sigmaVal=sign(pVal) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1341 if(is.na(pVal) || pVal==1 || pVal==0){ |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1342 returnColor = "#FFFFFF"; |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1343 }else{ |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1344 colVal=abs(pVal); |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1345 |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1346 if(sigmaVal<0){ |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1347 if(colVal>0.1) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1348 returnColor = "#CCFFCC"; |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1349 if(colVal<=0.1) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1350 returnColor = "#99FF99"; |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1351 if(colVal<=0.050) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1352 returnColor = "#66FF66"; |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1353 if(colVal<=0.010) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1354 returnColor = "#33FF33"; |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1355 if(colVal<=0.005) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1356 returnColor = "#00FF00"; |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1357 |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1358 }else{ |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1359 if(colVal>0.1) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1360 returnColor = "#FFCCCC"; |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1361 if(colVal<=0.1) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1362 returnColor = "#FF9999"; |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1363 if(colVal<=0.05) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1364 returnColor = "#FF6666"; |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1365 if(colVal<=0.01) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1366 returnColor = "#FF3333"; |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1367 if(colVal<0.005) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1368 returnColor = "#FF0000"; |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1369 } |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1370 } |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1371 |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1372 return(returnColor) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1373 } |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1374 |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1375 |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1376 |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1377 plotHelp <- function(xfrac=0.05,yfrac=0.05,log=FALSE){ |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1378 if(!log){ |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1379 x = par()$usr[1]-(par()$usr[2]-par()$usr[1])*xfrac |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1380 y = par()$usr[4]+(par()$usr[4]-par()$usr[3])*yfrac |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1381 }else { |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1382 if(log==2){ |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1383 x = par()$usr[1]-(par()$usr[2]-par()$usr[1])*xfrac |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1384 y = 10^((par()$usr[4])+((par()$usr[4])-(par()$usr[3]))*yfrac) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1385 } |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1386 if(log==1){ |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1387 x = 10^((par()$usr[1])-((par()$usr[2])-(par()$usr[1]))*xfrac) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1388 y = par()$usr[4]+(par()$usr[4]-par()$usr[3])*yfrac |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1389 } |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1390 if(log==3){ |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1391 x = 10^((par()$usr[1])-((par()$usr[2])-(par()$usr[1]))*xfrac) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1392 y = 10^((par()$usr[4])+((par()$usr[4])-(par()$usr[3]))*yfrac) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1393 } |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1394 } |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1395 return(c("x"=x,"y"=y)) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1396 } |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1397 |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1398 # SHMulation |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1399 |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1400 # Based on targeting, introduce a single mutation & then update the targeting |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1401 oneMutation <- function(){ |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1402 # Pick a postion + mutation |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1403 posMutation = sample(1:(seqGermlineLen*4),1,replace=F,prob=as.vector(seqTargeting)) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1404 posNucNumb = ceiling(posMutation/4) # Nucleotide number |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1405 posNucKind = 4 - ( (posNucNumb*4) - posMutation ) # Nuc the position mutates to |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1406 |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1407 #mutate the simulation sequence |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1408 seqSimVec <- s2c(seqSim) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1409 seqSimVec[posNucNumb] <- NUCLEOTIDES[posNucKind] |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1410 seqSim <<- c2s(seqSimVec) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1411 |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1412 #update Mutability, Targeting & MutationsTypes |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1413 updateMutabilityNTargeting(posNucNumb) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1414 |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1415 #return(c(posNucNumb,NUCLEOTIDES[posNucKind])) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1416 return(posNucNumb) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1417 } |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1418 |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1419 updateMutabilityNTargeting <- function(position){ |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1420 min_i<-max((position-2),1) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1421 max_i<-min((position+2),nchar(seqSim)) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1422 min_ii<-min(min_i,3) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1423 |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1424 #mutability - update locally |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1425 seqMutability[(min_i):(max_i)] <<- computeMutabilities(substr(seqSim,position-4,position+4))[(min_ii):(max_i-min_i+min_ii)] |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1426 |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1427 |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1428 #targeting - compute locally |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1429 seqTargeting[,min_i:max_i] <<- computeTargeting(substr(seqSim,min_i,max_i),seqMutability[min_i:max_i]) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1430 seqTargeting[is.na(seqTargeting)] <<- 0 |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1431 #mutCodonPos = getCodonPos(position) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1432 mutCodonPos = seq(getCodonPos(min_i)[1],getCodonPos(max_i)[3]) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1433 #cat(mutCodonPos,"\n") |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1434 mutTypeCodon = getCodonPos(position) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1435 seqMutationTypes[,mutTypeCodon] <<- computeMutationTypesFast( substr(seqSim,mutTypeCodon[1],mutTypeCodon[3]) ) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1436 # Stop = 0 |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1437 if(any(seqMutationTypes[,mutCodonPos]=="Stop",na.rm=T )){ |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1438 seqTargeting[,mutCodonPos][seqMutationTypes[,mutCodonPos]=="Stop"] <<- 0 |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1439 } |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1440 |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1441 |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1442 #Selection |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1443 selectedPos = (min_i*4-4)+(which(seqMutationTypes[,min_i:max_i]=="R")) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1444 # CDR |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1445 selectedCDR = selectedPos[which(matCDR[selectedPos]==T)] |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1446 seqTargeting[selectedCDR] <<- seqTargeting[selectedCDR] * exp(selCDR) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1447 seqTargeting[selectedCDR] <<- seqTargeting[selectedCDR]/baseLineCDR_K |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1448 |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1449 # FWR |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1450 selectedFWR = selectedPos[which(matFWR[selectedPos]==T)] |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1451 seqTargeting[selectedFWR] <<- seqTargeting[selectedFWR] * exp(selFWR) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1452 seqTargeting[selectedFWR] <<- seqTargeting[selectedFWR]/baseLineFWR_K |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1453 |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1454 } |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1455 |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1456 |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1457 |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1458 # Validate the mutation: if the mutation has not been sampled before validate it, else discard it. |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1459 validateMutation <- function(){ |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1460 if( !(mutatedPos%in%mutatedPositions) ){ # if it's a new mutation |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1461 uniqueMutationsIntroduced <<- uniqueMutationsIntroduced + 1 |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1462 mutatedPositions[uniqueMutationsIntroduced] <<- mutatedPos |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1463 }else{ |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1464 if(substr(seqSim,mutatedPos,mutatedPos)==substr(seqGermline,mutatedPos,mutatedPos)){ # back to germline mutation |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1465 mutatedPositions <<- mutatedPositions[-which(mutatedPositions==mutatedPos)] |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1466 uniqueMutationsIntroduced <<- uniqueMutationsIntroduced - 1 |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1467 } |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1468 } |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1469 } |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1470 |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1471 |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1472 |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1473 # Places text (labels) at normalized coordinates |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1474 myaxis <- function(xfrac=0.05,yfrac=0.05,log=FALSE,w="text",cex=1,adj=1,thecol="black"){ |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1475 par(xpd=TRUE) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1476 if(!log) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1477 text(par()$usr[1]-(par()$usr[2]-par()$usr[1])*xfrac,par()$usr[4]+(par()$usr[4]-par()$usr[3])*yfrac,w,cex=cex,adj=adj,col=thecol) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1478 else { |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1479 if(log==2) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1480 text( |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1481 par()$usr[1]-(par()$usr[2]-par()$usr[1])*xfrac, |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1482 10^((par()$usr[4])+((par()$usr[4])-(par()$usr[3]))*yfrac), |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1483 w,cex=cex,adj=adj,col=thecol) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1484 if(log==1) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1485 text( |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1486 10^((par()$usr[1])-((par()$usr[2])-(par()$usr[1]))*xfrac), |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1487 par()$usr[4]+(par()$usr[4]-par()$usr[3])*yfrac, |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1488 w,cex=cex,adj=adj,col=thecol) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1489 if(log==3) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1490 text( |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1491 10^((par()$usr[1])-((par()$usr[2])-(par()$usr[1]))*xfrac), |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1492 10^((par()$usr[4])+((par()$usr[4])-(par()$usr[3]))*yfrac), |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1493 w,cex=cex,adj=adj,col=thecol) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1494 } |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1495 par(xpd=FALSE) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1496 } |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1497 |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1498 |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1499 |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1500 # Count the mutations in a sequence |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1501 analyzeMutations <- function( inputMatrixIndex, model = 0 , multipleMutation=0, seqWithStops=0){ |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1502 |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1503 paramGL = s2c(matInput[inputMatrixIndex,2]) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1504 paramSeq = s2c(matInput[inputMatrixIndex,1]) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1505 |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1506 #if( any(paramSeq=="N") ){ |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1507 # gapPos_Seq = which(paramSeq=="N") |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1508 # gapPos_Seq_ToReplace = gapPos_Seq[paramGL[gapPos_Seq] != "N"] |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1509 # paramSeq[gapPos_Seq_ToReplace] = paramGL[gapPos_Seq_ToReplace] |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1510 #} |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1511 mutations_val = paramGL != paramSeq |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1512 |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1513 if(any(mutations_val)){ |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1514 mutationPos = which(mutations_val)#{1:length(mutations_val)}[mutations_val] |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1515 length_mutations =length(mutationPos) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1516 mutationInfo = rep(NA,length_mutations) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1517 |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1518 pos<- mutationPos |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1519 pos_array<-array(sapply(pos,getCodonPos)) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1520 codonGL = paramGL[pos_array] |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1521 codonSeqWhole = paramSeq[pos_array] |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1522 codonSeq = sapply(pos,function(x){ |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1523 seqP = paramGL[getCodonPos(x)] |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1524 muCodonPos = {x-1}%%3+1 |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1525 seqP[muCodonPos] = paramSeq[x] |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1526 return(seqP) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1527 }) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1528 GLcodons = apply(matrix(codonGL,length_mutations,3,byrow=TRUE),1,c2s) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1529 SeqcodonsWhole = apply(matrix(codonSeqWhole,length_mutations,3,byrow=TRUE),1,c2s) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1530 Seqcodons = apply(codonSeq,2,c2s) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1531 |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1532 mutationInfo = apply(rbind(GLcodons , Seqcodons),2,function(x){mutationType(c2s(x[1]),c2s(x[2]))}) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1533 names(mutationInfo) = mutationPos |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1534 |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1535 mutationInfoWhole = apply(rbind(GLcodons , SeqcodonsWhole),2,function(x){mutationType(c2s(x[1]),c2s(x[2]))}) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1536 names(mutationInfoWhole) = mutationPos |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1537 |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1538 mutationInfo <- mutationInfo[!is.na(mutationInfo)] |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1539 mutationInfoWhole <- mutationInfoWhole[!is.na(mutationInfoWhole)] |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1540 |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1541 if(any(!is.na(mutationInfo))){ |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1542 |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1543 #Filter based on Stop (at the codon level) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1544 if(seqWithStops==1){ |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1545 nucleotidesAtStopCodons = names(mutationInfoWhole[mutationInfoWhole!="Stop"]) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1546 mutationInfo = mutationInfo[nucleotidesAtStopCodons] |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1547 mutationInfoWhole = mutationInfo[nucleotidesAtStopCodons] |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1548 }else{ |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1549 countStops = sum(mutationInfoWhole=="Stop") |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1550 if(seqWithStops==2 & countStops==0) mutationInfo = NA |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1551 if(seqWithStops==3 & countStops>0) mutationInfo = NA |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1552 } |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1553 |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1554 if(any(!is.na(mutationInfo))){ |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1555 #Filter mutations based on multipleMutation |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1556 if(multipleMutation==1 & !is.na(mutationInfo)){ |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1557 mutationCodons = getCodonNumb(as.numeric(names(mutationInfoWhole))) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1558 tableMutationCodons <- table(mutationCodons) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1559 codonsWithMultipleMutations <- as.numeric(names(tableMutationCodons[tableMutationCodons>1])) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1560 if(any(codonsWithMultipleMutations)){ |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1561 #remove the nucleotide mutations in the codons with multiple mutations |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1562 mutationInfo <- mutationInfo[!(mutationCodons %in% codonsWithMultipleMutations)] |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1563 #replace those codons with Ns in the input sequence |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1564 paramSeq[unlist(lapply(codonsWithMultipleMutations, getCodonNucs))] = "N" |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1565 matInput[inputMatrixIndex,1] <<- c2s(paramSeq) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1566 } |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1567 } |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1568 |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1569 #Filter mutations based on the model |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1570 if(any(mutationInfo)==T | is.na(any(mutationInfo))){ |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1571 |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1572 if(model==1 & !is.na(mutationInfo)){ |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1573 mutationInfo <- mutationInfo[mutationInfo=="S"] |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1574 } |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1575 if(any(mutationInfo)==T | is.na(any(mutationInfo))) return(mutationInfo) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1576 else return(NA) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1577 }else{ |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1578 return(NA) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1579 } |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1580 }else{ |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1581 return(NA) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1582 } |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1583 |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1584 |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1585 }else{ |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1586 return(NA) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1587 } |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1588 |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1589 |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1590 }else{ |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1591 return (NA) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1592 } |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1593 } |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1594 |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1595 analyzeMutationsFixed <- function( inputArray, model = 0 , multipleMutation=0, seqWithStops=0){ |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1596 |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1597 paramGL = s2c(inputArray[2]) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1598 paramSeq = s2c(inputArray[1]) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1599 inputSeq <- inputArray[1] |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1600 #if( any(paramSeq=="N") ){ |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1601 # gapPos_Seq = which(paramSeq=="N") |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1602 # gapPos_Seq_ToReplace = gapPos_Seq[paramGL[gapPos_Seq] != "N"] |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1603 # paramSeq[gapPos_Seq_ToReplace] = paramGL[gapPos_Seq_ToReplace] |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1604 #} |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1605 mutations_val = paramGL != paramSeq |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1606 |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1607 if(any(mutations_val)){ |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1608 mutationPos = which(mutations_val)#{1:length(mutations_val)}[mutations_val] |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1609 length_mutations =length(mutationPos) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1610 mutationInfo = rep(NA,length_mutations) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1611 |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1612 pos<- mutationPos |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1613 pos_array<-array(sapply(pos,getCodonPos)) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1614 codonGL = paramGL[pos_array] |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1615 codonSeqWhole = paramSeq[pos_array] |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1616 codonSeq = sapply(pos,function(x){ |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1617 seqP = paramGL[getCodonPos(x)] |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1618 muCodonPos = {x-1}%%3+1 |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1619 seqP[muCodonPos] = paramSeq[x] |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1620 return(seqP) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1621 }) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1622 GLcodons = apply(matrix(codonGL,length_mutations,3,byrow=TRUE),1,c2s) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1623 SeqcodonsWhole = apply(matrix(codonSeqWhole,length_mutations,3,byrow=TRUE),1,c2s) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1624 Seqcodons = apply(codonSeq,2,c2s) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1625 |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1626 mutationInfo = apply(rbind(GLcodons , Seqcodons),2,function(x){mutationType(c2s(x[1]),c2s(x[2]))}) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1627 names(mutationInfo) = mutationPos |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1628 |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1629 mutationInfoWhole = apply(rbind(GLcodons , SeqcodonsWhole),2,function(x){mutationType(c2s(x[1]),c2s(x[2]))}) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1630 names(mutationInfoWhole) = mutationPos |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1631 |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1632 mutationInfo <- mutationInfo[!is.na(mutationInfo)] |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1633 mutationInfoWhole <- mutationInfoWhole[!is.na(mutationInfoWhole)] |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1634 |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1635 if(any(!is.na(mutationInfo))){ |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1636 |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1637 #Filter based on Stop (at the codon level) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1638 if(seqWithStops==1){ |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1639 nucleotidesAtStopCodons = names(mutationInfoWhole[mutationInfoWhole!="Stop"]) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1640 mutationInfo = mutationInfo[nucleotidesAtStopCodons] |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1641 mutationInfoWhole = mutationInfo[nucleotidesAtStopCodons] |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1642 }else{ |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1643 countStops = sum(mutationInfoWhole=="Stop") |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1644 if(seqWithStops==2 & countStops==0) mutationInfo = NA |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1645 if(seqWithStops==3 & countStops>0) mutationInfo = NA |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1646 } |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1647 |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1648 if(any(!is.na(mutationInfo))){ |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1649 #Filter mutations based on multipleMutation |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1650 if(multipleMutation==1 & !is.na(mutationInfo)){ |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1651 mutationCodons = getCodonNumb(as.numeric(names(mutationInfoWhole))) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1652 tableMutationCodons <- table(mutationCodons) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1653 codonsWithMultipleMutations <- as.numeric(names(tableMutationCodons[tableMutationCodons>1])) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1654 if(any(codonsWithMultipleMutations)){ |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1655 #remove the nucleotide mutations in the codons with multiple mutations |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1656 mutationInfo <- mutationInfo[!(mutationCodons %in% codonsWithMultipleMutations)] |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1657 #replace those codons with Ns in the input sequence |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1658 paramSeq[unlist(lapply(codonsWithMultipleMutations, getCodonNucs))] = "N" |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1659 #matInput[inputMatrixIndex,1] <<- c2s(paramSeq) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1660 inputSeq <- c2s(paramSeq) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1661 } |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1662 } |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1663 |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1664 #Filter mutations based on the model |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1665 if(any(mutationInfo)==T | is.na(any(mutationInfo))){ |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1666 |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1667 if(model==1 & !is.na(mutationInfo)){ |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1668 mutationInfo <- mutationInfo[mutationInfo=="S"] |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1669 } |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1670 if(any(mutationInfo)==T | is.na(any(mutationInfo))) return(list(mutationInfo,inputSeq)) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1671 else return(list(NA,inputSeq)) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1672 }else{ |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1673 return(list(NA,inputSeq)) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1674 } |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1675 }else{ |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1676 return(list(NA,inputSeq)) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1677 } |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1678 |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1679 |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1680 }else{ |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1681 return(list(NA,inputSeq)) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1682 } |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1683 |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1684 |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1685 }else{ |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1686 return (list(NA,inputSeq)) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1687 } |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1688 } |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1689 |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1690 # triMutability Background Count |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1691 buildMutabilityModel <- function( inputMatrixIndex, model=0 , multipleMutation=0, seqWithStops=0, stopMutations=0){ |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1692 |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1693 #rowOrigMatInput = matInput[inputMatrixIndex,] |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1694 seqGL = gsub("-", "", matInput[inputMatrixIndex,2]) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1695 seqInput = gsub("-", "", matInput[inputMatrixIndex,1]) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1696 #matInput[inputMatrixIndex,] <<- cbind(seqInput,seqGL) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1697 tempInput <- cbind(seqInput,seqGL) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1698 seqLength = nchar(seqGL) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1699 list_analyzeMutationsFixed<- analyzeMutationsFixed(tempInput, model, multipleMutation, seqWithStops) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1700 mutationCount <- list_analyzeMutationsFixed[[1]] |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1701 seqInput <- list_analyzeMutationsFixed[[2]] |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1702 BackgroundMatrix = mutabilityMatrix |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1703 MutationMatrix = mutabilityMatrix |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1704 MutationCountMatrix = mutabilityMatrix |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1705 if(!is.na(mutationCount)){ |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1706 if((stopMutations==0 & model==0) | (stopMutations==1 & (sum(mutationCount=="Stop")<length(mutationCount))) | (model==1 & (sum(mutationCount=="S")>0)) ){ |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1707 |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1708 fivermerStartPos = 1:(seqLength-4) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1709 fivemerLength <- length(fivermerStartPos) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1710 fivemerGL <- substr(rep(seqGL,length(fivermerStartPos)),(fivermerStartPos),(fivermerStartPos+4)) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1711 fivemerSeq <- substr(rep(seqInput,length(fivermerStartPos)),(fivermerStartPos),(fivermerStartPos+4)) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1712 |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1713 #Background |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1714 for(fivemerIndex in 1:fivemerLength){ |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1715 fivemer = fivemerGL[fivemerIndex] |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1716 if(!any(grep("N",fivemer))){ |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1717 fivemerCodonPos = fivemerCodon(fivemerIndex) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1718 fivemerReadingFrameCodon = substr(fivemer,fivemerCodonPos[1],fivemerCodonPos[3]) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1719 fivemerReadingFrameCodonInputSeq = substr(fivemerSeq[fivemerIndex],fivemerCodonPos[1],fivemerCodonPos[3]) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1720 |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1721 # All mutations model |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1722 #if(!any(grep("N",fivemerReadingFrameCodon))){ |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1723 if(model==0){ |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1724 if(stopMutations==0){ |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1725 if(!any(grep("N",fivemerReadingFrameCodonInputSeq))) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1726 BackgroundMatrix[fivemer] <- (BackgroundMatrix[fivemer] + 1) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1727 }else{ |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1728 if( !any(grep("N",fivemerReadingFrameCodonInputSeq)) & translateCodonToAminoAcid(fivemerReadingFrameCodon)!="*" ){ |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1729 positionWithinCodon = which(fivemerCodonPos==3)#positionsWithinCodon[(fivemerCodonPos[1]%%3)+1] |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1730 BackgroundMatrix[fivemer] <- (BackgroundMatrix[fivemer] + probNonStopMutations[fivemerReadingFrameCodon,positionWithinCodon]) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1731 } |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1732 } |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1733 }else{ # Only silent mutations |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1734 if( !any(grep("N",fivemerReadingFrameCodonInputSeq)) & translateCodonToAminoAcid(fivemerReadingFrameCodon)!="*" & translateCodonToAminoAcid(fivemerReadingFrameCodonInputSeq)==translateCodonToAminoAcid(fivemerReadingFrameCodon)){ |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1735 positionWithinCodon = which(fivemerCodonPos==3) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1736 BackgroundMatrix[fivemer] <- (BackgroundMatrix[fivemer] + probSMutations[fivemerReadingFrameCodon,positionWithinCodon]) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1737 } |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1738 } |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1739 #} |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1740 } |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1741 } |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1742 |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1743 #Mutations |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1744 if(stopMutations==1) mutationCount = mutationCount[mutationCount!="Stop"] |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1745 if(model==1) mutationCount = mutationCount[mutationCount=="S"] |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1746 mutationPositions = as.numeric(names(mutationCount)) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1747 mutationCount = mutationCount[mutationPositions>2 & mutationPositions<(seqLength-1)] |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1748 mutationPositions = mutationPositions[mutationPositions>2 & mutationPositions<(seqLength-1)] |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1749 countMutations = 0 |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1750 for(mutationPosition in mutationPositions){ |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1751 fivemerIndex = mutationPosition-2 |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1752 fivemer = fivemerSeq[fivemerIndex] |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1753 GLfivemer = fivemerGL[fivemerIndex] |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1754 fivemerCodonPos = fivemerCodon(fivemerIndex) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1755 fivemerReadingFrameCodon = substr(fivemer,fivemerCodonPos[1],fivemerCodonPos[3]) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1756 fivemerReadingFrameCodonGL = substr(GLfivemer,fivemerCodonPos[1],fivemerCodonPos[3]) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1757 if(!any(grep("N",fivemer)) & !any(grep("N",GLfivemer))){ |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1758 if(model==0){ |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1759 countMutations = countMutations + 1 |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1760 MutationMatrix[GLfivemer] <- (MutationMatrix[GLfivemer] + 1) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1761 MutationCountMatrix[GLfivemer] <- (MutationCountMatrix[GLfivemer] + 1) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1762 }else{ |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1763 if( translateCodonToAminoAcid(fivemerReadingFrameCodonGL)!="*" ){ |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1764 countMutations = countMutations + 1 |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1765 positionWithinCodon = which(fivemerCodonPos==3) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1766 glNuc = substr(fivemerReadingFrameCodonGL,positionWithinCodon,positionWithinCodon) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1767 inputNuc = substr(fivemerReadingFrameCodon,positionWithinCodon,positionWithinCodon) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1768 MutationMatrix[GLfivemer] <- (MutationMatrix[GLfivemer] + substitution[glNuc,inputNuc]) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1769 MutationCountMatrix[GLfivemer] <- (MutationCountMatrix[GLfivemer] + 1) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1770 } |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1771 } |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1772 } |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1773 } |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1774 |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1775 seqMutability = MutationMatrix/BackgroundMatrix |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1776 seqMutability = seqMutability/sum(seqMutability,na.rm=TRUE) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1777 #cat(inputMatrixIndex,"\t",countMutations,"\n") |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1778 return(list("seqMutability" = seqMutability,"numbMutations" = countMutations,"seqMutabilityCount" = MutationCountMatrix, "BackgroundMatrix"=BackgroundMatrix)) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1779 |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1780 } |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1781 } |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1782 |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1783 } |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1784 |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1785 #Returns the codon position containing the middle nucleotide |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1786 fivemerCodon <- function(fivemerIndex){ |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1787 codonPos = list(2:4,1:3,3:5) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1788 fivemerType = fivemerIndex%%3 |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1789 return(codonPos[[fivemerType+1]]) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1790 } |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1791 |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1792 #returns probability values for one mutation in codons resulting in R, S or Stop |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1793 probMutations <- function(typeOfMutation){ |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1794 matMutationProb <- matrix(0,ncol=3,nrow=125,dimnames=list(words(alphabet = c(NUCLEOTIDES,"N"), length=3),c(1:3))) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1795 for(codon in rownames(matMutationProb)){ |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1796 if( !any(grep("N",codon)) ){ |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1797 for(muPos in 1:3){ |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1798 matCodon = matrix(rep(s2c(codon),3),nrow=3,ncol=3,byrow=T) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1799 glNuc = matCodon[1,muPos] |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1800 matCodon[,muPos] = canMutateTo(glNuc) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1801 substitutionRate = substitution[glNuc,matCodon[,muPos]] |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1802 typeOfMutations = apply(rbind(rep(codon,3),apply(matCodon,1,c2s)),2,function(x){mutationType(c2s(x[1]),c2s(x[2]))}) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1803 matMutationProb[codon,muPos] <- sum(substitutionRate[typeOfMutations==typeOfMutation]) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1804 } |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1805 } |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1806 } |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1807 |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1808 return(matMutationProb) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1809 } |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1810 |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1811 |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1812 |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1813 |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1814 #Mapping Trinucleotides to fivemers |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1815 mapTriToFivemer <- function(triMutability=triMutability_Literature_Human){ |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1816 rownames(triMutability) <- triMutability_Names |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1817 Fivemer<-rep(NA,1024) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1818 names(Fivemer)<-words(alphabet=NUCLEOTIDES,length=5) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1819 Fivemer<-sapply(names(Fivemer),function(Word)return(sum( c(triMutability[substring(Word,3,5),1],triMutability[substring(Word,2,4),2],triMutability[substring(Word,1,3),3]),na.rm=TRUE))) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1820 Fivemer<-Fivemer/sum(Fivemer) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1821 return(Fivemer) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1822 } |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1823 |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1824 collapseFivemerToTri<-function(Fivemer,Weights=MutabilityWeights,position=1,NUC="A"){ |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1825 Indices<-substring(names(Fivemer),3,3)==NUC |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1826 Factors<-substring(names(Fivemer[Indices]),(4-position),(6-position)) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1827 tapply(which(Indices),Factors,function(i)weighted.mean(Fivemer[i],Weights[i],na.rm=TRUE)) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1828 } |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1829 |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1830 |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1831 |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1832 CountFivemerToTri<-function(Fivemer,Weights=MutabilityWeights,position=1,NUC="A"){ |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1833 Indices<-substring(names(Fivemer),3,3)==NUC |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1834 Factors<-substring(names(Fivemer[Indices]),(4-position),(6-position)) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1835 tapply(which(Indices),Factors,function(i)sum(Weights[i],na.rm=TRUE)) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1836 } |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1837 |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1838 #Uses the real counts of the mutated fivemers |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1839 CountFivemerToTri2<-function(Fivemer,Counts=MutabilityCounts,position=1,NUC="A"){ |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1840 Indices<-substring(names(Fivemer),3,3)==NUC |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1841 Factors<-substring(names(Fivemer[Indices]),(4-position),(6-position)) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1842 tapply(which(Indices),Factors,function(i)sum(Counts[i],na.rm=TRUE)) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1843 } |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1844 |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1845 bootstrap<-function(x=c(33,12,21),M=10000,alpha=0.05){ |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1846 N<-sum(x) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1847 if(N){ |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1848 p<-x/N |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1849 k<-length(x)-1 |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1850 tmp<-rmultinom(M, size = N, prob=p) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1851 tmp_p<-apply(tmp,2,function(y)y/N) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1852 (apply(tmp_p,1,function(y)quantile(y,c(alpha/2/k,1-alpha/2/k)))) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1853 } |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1854 else return(matrix(0,2,length(x))) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1855 } |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1856 |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1857 |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1858 |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1859 |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1860 bootstrap2<-function(x=c(33,12,21),n=10,M=10000,alpha=0.05){ |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1861 |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1862 N<-sum(x) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1863 k<-length(x) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1864 y<-rep(1:k,x) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1865 tmp<-sapply(1:M,function(i)sample(y,n)) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1866 if(n>1)tmp_p<-sapply(1:M,function(j)sapply(1:k,function(i)sum(tmp[,j]==i)))/n |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1867 if(n==1)tmp_p<-sapply(1:M,function(j)sapply(1:k,function(i)sum(tmp[j]==i)))/n |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1868 (apply(tmp_p,1,function(z)quantile(z,c(alpha/2/(k-1),1-alpha/2/(k-1))))) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1869 } |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1870 |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1871 |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1872 |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1873 p_value<-function(x=c(33,12,21),M=100000,x_obs=c(2,5,3)){ |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1874 n=sum(x_obs) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1875 N<-sum(x) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1876 k<-length(x) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1877 y<-rep(1:k,x) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1878 tmp<-sapply(1:M,function(i)sample(y,n)) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1879 if(n>1)tmp_p<-sapply(1:M,function(j)sapply(1:k,function(i)sum(tmp[,j]==i))) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1880 if(n==1)tmp_p<-sapply(1:M,function(j)sapply(1:k,function(i)sum(tmp[j]==i))) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1881 tmp<-rbind(sapply(1:3,function(i)sum(tmp_p[i,]>=x_obs[i])/M), |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1882 sapply(1:3,function(i)sum(tmp_p[i,]<=x_obs[i])/M)) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1883 sapply(1:3,function(i){if(tmp[1,i]>=tmp[2,i])return(-tmp[2,i])else return(tmp[1,i])}) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1884 } |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1885 |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1886 #"D:\\Sequences\\IMGT Germlines\\Human_SNPless_IGHJ.FASTA" |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1887 # Remove SNPs from IMGT germline segment alleles |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1888 generateUnambiguousRepertoire <- function(repertoireInFile,repertoireOutFile){ |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1889 repertoireIn <- read.fasta(repertoireInFile, seqtype="DNA",as.string=T,set.attributes=F,forceDNAtolower=F) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1890 alleleNames <- sapply(names(repertoireIn),function(x)strsplit(x,"|",fixed=TRUE)[[1]][2]) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1891 SNPs <- tapply(repertoireIn,sapply(alleleNames,function(x)strsplit(x,"*",fixed=TRUE)[[1]][1]),function(x){ |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1892 Indices<-NULL |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1893 for(i in 1:length(x)){ |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1894 firstSeq = s2c(x[[1]]) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1895 iSeq = s2c(x[[i]]) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1896 Indices<-c(Indices,which(firstSeq[1:320]!=iSeq[1:320] & firstSeq[1:320]!="." & iSeq[1:320]!="." )) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1897 } |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1898 return(sort(unique(Indices))) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1899 }) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1900 repertoireOut <- repertoireIn |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1901 repertoireOut <- lapply(names(repertoireOut), function(repertoireName){ |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1902 alleleName <- strsplit(repertoireName,"|",fixed=TRUE)[[1]][2] |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1903 geneSegmentName <- strsplit(alleleName,"*",fixed=TRUE)[[1]][1] |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1904 alleleSeq <- s2c(repertoireOut[[repertoireName]]) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1905 alleleSeq[as.numeric(unlist(SNPs[geneSegmentName]))] <- "N" |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1906 alleleSeq <- c2s(alleleSeq) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1907 repertoireOut[[repertoireName]] <- alleleSeq |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1908 }) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1909 names(repertoireOut) <- names(repertoireIn) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1910 write.fasta(repertoireOut,names(repertoireOut),file.out=repertoireOutFile) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1911 |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1912 } |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1913 |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1914 |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1915 |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1916 |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1917 |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1918 |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1919 ############ |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1920 groupBayes2 = function(indexes, param_resultMat){ |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1921 |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1922 BayesGDist_Focused_CDR = calculate_bayesG( x=param_resultMat[indexes,1], N=apply(param_resultMat[indexes,c(1,2,4)],1,sum,na.rm=T), p=apply(param_resultMat[indexes,5:8],1,function(x){x[1]/(x[1]+x[2]+x[4])})) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1923 BayesGDist_Focused_FWR = calculate_bayesG( x=param_resultMat[indexes,3], N=apply(param_resultMat[indexes,c(3,2,4)],1,sum,na.rm=T), p=apply(param_resultMat[indexes,5:8],1,function(x){x[3]/(x[3]+x[2]+x[4])})) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1924 #BayesGDist_Local_CDR = calculate_bayesG( x=param_resultMat[indexes,1], N=apply(param_resultMat[indexes,c(1,2)],1,sum,na.rm=T), p=apply(param_resultMat[indexes,5:8],1,function(x){x[1]/(x[1]+x[2])})) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1925 #BayesGDist_Local_FWR = calculate_bayesG( x=param_resultMat[indexes,3], N=apply(param_resultMat[indexes,c(3,4)],1,sum,na.rm=T), p=apply(param_resultMat[indexes,5:8],1,function(x){x[3]/(x[3]+x[4])})) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1926 #BayesGDist_Global_CDR = calculate_bayesG( x=param_resultMat[indexes,1], N=apply(param_resultMat[indexes,c(1,2,3,4)],1,sum,na.rm=T), p=apply(param_resultMat[indexes,5:8],1,function(x){x[1]/(x[1]+x[2]+x[3]+x[4])})) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1927 #BayesGDist_Global_FWR = calculate_bayesG( x=param_resultMat[indexes,3], N=apply(param_resultMat[indexes,c(1,2,3,4)],1,sum,na.rm=T), p=apply(param_resultMat[indexes,5:8],1,function(x){x[3]/(x[1]+x[2]+x[3]+x[4])})) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1928 return ( list("BayesGDist_Focused_CDR"=BayesGDist_Focused_CDR, |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1929 "BayesGDist_Focused_FWR"=BayesGDist_Focused_FWR) ) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1930 #"BayesGDist_Local_CDR"=BayesGDist_Local_CDR, |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1931 #"BayesGDist_Local_FWR" = BayesGDist_Local_FWR)) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1932 # "BayesGDist_Global_CDR" = BayesGDist_Global_CDR, |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1933 # "BayesGDist_Global_FWR" = BayesGDist_Global_FWR) ) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1934 |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1935 |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1936 } |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1937 |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1938 |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1939 calculate_bayesG <- function( x=array(), N=array(), p=array(), max_sigma=20, length_sigma=4001){ |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1940 G <- max(length(x),length(N),length(p)) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1941 x=array(x,dim=G) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1942 N=array(N,dim=G) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1943 p=array(p,dim=G) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1944 |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1945 indexOfZero = N>0 & p>0 |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1946 N = N[indexOfZero] |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1947 x = x[indexOfZero] |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1948 p = p[indexOfZero] |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1949 G <- length(x) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1950 |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1951 if(G){ |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1952 |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1953 cons<-array( dim=c(length_sigma,G) ) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1954 if(G==1) { |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1955 return(calculate_bayes(x=x[G],N=N[G],p=p[G],max_sigma=max_sigma,length_sigma=length_sigma)) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1956 } |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1957 else { |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1958 for(g in 1:G) cons[,g] <- calculate_bayes(x=x[g],N=N[g],p=p[g],max_sigma=max_sigma,length_sigma=length_sigma) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1959 listMatG <- convolutionPowersOfTwoByTwos(cons,length_sigma=length_sigma) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1960 y<-calculate_bayesGHelper(listMatG,length_sigma=length_sigma) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1961 return( y/sum(y)/(2*max_sigma/(length_sigma-1)) ) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1962 } |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1963 }else{ |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1964 return(NA) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1965 } |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1966 } |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1967 |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1968 |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1969 calculate_bayesGHelper <- function( listMatG,length_sigma=4001 ){ |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1970 matG <- listMatG[[1]] |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1971 groups <- listMatG[[2]] |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1972 i = 1 |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1973 resConv <- matG[,i] |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1974 denom <- 2^groups[i] |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1975 if(length(groups)>1){ |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1976 while( i<length(groups) ){ |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1977 i = i + 1 |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1978 resConv <- weighted_conv(resConv, matG[,i], w= {{2^groups[i]}/denom} ,length_sigma=length_sigma) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1979 #cat({{2^groups[i]}/denom},"\n") |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1980 denom <- denom + 2^groups[i] |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1981 } |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1982 } |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1983 return(resConv) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1984 } |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1985 |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1986 weighted_conv<-function(x,y,w=1,m=100,length_sigma=4001){ |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1987 lx<-length(x) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1988 ly<-length(y) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1989 if({lx<m}| {{lx*w}<m}| {{ly}<m}| {{ly*w}<m}){ |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1990 if(w<1){ |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1991 y1<-approx(1:ly,y,seq(1,ly,length.out=m))$y |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1992 x1<-approx(1:lx,x,seq(1,lx,length.out=m/w))$y |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1993 lx<-length(x1) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1994 ly<-length(y1) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1995 } |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1996 else { |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1997 y1<-approx(1:ly,y,seq(1,ly,length.out=m*w))$y |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1998 x1<-approx(1:lx,x,seq(1,lx,length.out=m))$y |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
1999 lx<-length(x1) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
2000 ly<-length(y1) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
2001 } |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
2002 } |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
2003 else{ |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
2004 x1<-x |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
2005 y1<-approx(1:ly,y,seq(1,ly,length.out=floor(lx*w)))$y |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
2006 ly<-length(y1) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
2007 } |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
2008 tmp<-approx(x=1:(lx+ly-1),y=convolve(x1,rev(y1),type="open"),xout=seq(1,lx+ly-1,length.out=length_sigma))$y |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
2009 tmp[tmp<=0] = 0 |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
2010 return(tmp/sum(tmp)) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
2011 } |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
2012 |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
2013 ######################## |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
2014 |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
2015 |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
2016 |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
2017 |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
2018 mutabilityMatrixONE<-rep(0,4) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
2019 names(mutabilityMatrixONE)<-NUCLEOTIDES |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
2020 |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
2021 # triMutability Background Count |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
2022 buildMutabilityModelONE <- function( inputMatrixIndex, model=0 , multipleMutation=0, seqWithStops=0, stopMutations=0){ |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
2023 |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
2024 #rowOrigMatInput = matInput[inputMatrixIndex,] |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
2025 seqGL = gsub("-", "", matInput[inputMatrixIndex,2]) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
2026 seqInput = gsub("-", "", matInput[inputMatrixIndex,1]) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
2027 matInput[inputMatrixIndex,] <<- c(seqInput,seqGL) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
2028 seqLength = nchar(seqGL) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
2029 mutationCount <- analyzeMutations(inputMatrixIndex, model, multipleMutation, seqWithStops) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
2030 BackgroundMatrix = mutabilityMatrixONE |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
2031 MutationMatrix = mutabilityMatrixONE |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
2032 MutationCountMatrix = mutabilityMatrixONE |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
2033 if(!is.na(mutationCount)){ |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
2034 if((stopMutations==0 & model==0) | (stopMutations==1 & (sum(mutationCount=="Stop")<length(mutationCount))) | (model==1 & (sum(mutationCount=="S")>0)) ){ |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
2035 |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
2036 # ONEmerStartPos = 1:(seqLength) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
2037 # ONEmerLength <- length(ONEmerStartPos) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
2038 ONEmerGL <- s2c(seqGL) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
2039 ONEmerSeq <- s2c(seqInput) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
2040 |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
2041 #Background |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
2042 for(ONEmerIndex in 1:seqLength){ |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
2043 ONEmer = ONEmerGL[ONEmerIndex] |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
2044 if(ONEmer!="N"){ |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
2045 ONEmerCodonPos = getCodonPos(ONEmerIndex) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
2046 ONEmerReadingFrameCodon = c2s(ONEmerGL[ONEmerCodonPos]) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
2047 ONEmerReadingFrameCodonInputSeq = c2s(ONEmerSeq[ONEmerCodonPos] ) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
2048 |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
2049 # All mutations model |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
2050 #if(!any(grep("N",ONEmerReadingFrameCodon))){ |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
2051 if(model==0){ |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
2052 if(stopMutations==0){ |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
2053 if(!any(grep("N",ONEmerReadingFrameCodonInputSeq))) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
2054 BackgroundMatrix[ONEmer] <- (BackgroundMatrix[ONEmer] + 1) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
2055 }else{ |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
2056 if( !any(grep("N",ONEmerReadingFrameCodonInputSeq)) & translateCodonToAminoAcid(ONEmerReadingFrameCodonInputSeq)!="*"){ |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
2057 positionWithinCodon = which(ONEmerCodonPos==ONEmerIndex)#positionsWithinCodon[(ONEmerCodonPos[1]%%3)+1] |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
2058 BackgroundMatrix[ONEmer] <- (BackgroundMatrix[ONEmer] + probNonStopMutations[ONEmerReadingFrameCodon,positionWithinCodon]) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
2059 } |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
2060 } |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
2061 }else{ # Only silent mutations |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
2062 if( !any(grep("N",ONEmerReadingFrameCodonInputSeq)) & translateCodonToAminoAcid(ONEmerReadingFrameCodonInputSeq)!="*" & translateCodonToAminoAcid(ONEmerReadingFrameCodonInputSeq)==translateCodonToAminoAcid(ONEmerReadingFrameCodon) ){ |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
2063 positionWithinCodon = which(ONEmerCodonPos==ONEmerIndex) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
2064 BackgroundMatrix[ONEmer] <- (BackgroundMatrix[ONEmer] + probSMutations[ONEmerReadingFrameCodon,positionWithinCodon]) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
2065 } |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
2066 } |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
2067 } |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
2068 } |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
2069 } |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
2070 |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
2071 #Mutations |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
2072 if(stopMutations==1) mutationCount = mutationCount[mutationCount!="Stop"] |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
2073 if(model==1) mutationCount = mutationCount[mutationCount=="S"] |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
2074 mutationPositions = as.numeric(names(mutationCount)) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
2075 mutationCount = mutationCount[mutationPositions>2 & mutationPositions<(seqLength-1)] |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
2076 mutationPositions = mutationPositions[mutationPositions>2 & mutationPositions<(seqLength-1)] |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
2077 countMutations = 0 |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
2078 for(mutationPosition in mutationPositions){ |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
2079 ONEmerIndex = mutationPosition |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
2080 ONEmer = ONEmerSeq[ONEmerIndex] |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
2081 GLONEmer = ONEmerGL[ONEmerIndex] |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
2082 ONEmerCodonPos = getCodonPos(ONEmerIndex) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
2083 ONEmerReadingFrameCodon = c2s(ONEmerSeq[ONEmerCodonPos]) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
2084 ONEmerReadingFrameCodonGL =c2s(ONEmerGL[ONEmerCodonPos]) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
2085 if(!any(grep("N",ONEmer)) & !any(grep("N",GLONEmer))){ |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
2086 if(model==0){ |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
2087 countMutations = countMutations + 1 |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
2088 MutationMatrix[GLONEmer] <- (MutationMatrix[GLONEmer] + 1) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
2089 MutationCountMatrix[GLONEmer] <- (MutationCountMatrix[GLONEmer] + 1) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
2090 }else{ |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
2091 if( translateCodonToAminoAcid(ONEmerReadingFrameCodonGL)!="*" ){ |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
2092 countMutations = countMutations + 1 |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
2093 positionWithinCodon = which(ONEmerCodonPos==ONEmerIndex) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
2094 glNuc = substr(ONEmerReadingFrameCodonGL,positionWithinCodon,positionWithinCodon) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
2095 inputNuc = substr(ONEmerReadingFrameCodon,positionWithinCodon,positionWithinCodon) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
2096 MutationMatrix[GLONEmer] <- (MutationMatrix[GLONEmer] + substitution[glNuc,inputNuc]) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
2097 MutationCountMatrix[GLONEmer] <- (MutationCountMatrix[GLONEmer] + 1) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
2098 } |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
2099 } |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
2100 } |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
2101 } |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
2102 |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
2103 seqMutability = MutationMatrix/BackgroundMatrix |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
2104 seqMutability = seqMutability/sum(seqMutability,na.rm=TRUE) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
2105 #cat(inputMatrixIndex,"\t",countMutations,"\n") |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
2106 return(list("seqMutability" = seqMutability,"numbMutations" = countMutations,"seqMutabilityCount" = MutationCountMatrix, "BackgroundMatrix"=BackgroundMatrix)) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
2107 # tmp<-list("seqMutability" = seqMutability,"numbMutations" = countMutations,"seqMutabilityCount" = MutationCountMatrix) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
2108 } |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
2109 } |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
2110 |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
2111 ################ |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
2112 # $Id: trim.R 989 2006-10-29 15:28:26Z ggorjan $ |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
2113 |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
2114 trim <- function(s, recode.factor=TRUE, ...) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
2115 UseMethod("trim", s) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
2116 |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
2117 trim.default <- function(s, recode.factor=TRUE, ...) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
2118 s |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
2119 |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
2120 trim.character <- function(s, recode.factor=TRUE, ...) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
2121 { |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
2122 s <- sub(pattern="^ +", replacement="", x=s) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
2123 s <- sub(pattern=" +$", replacement="", x=s) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
2124 s |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
2125 } |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
2126 |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
2127 trim.factor <- function(s, recode.factor=TRUE, ...) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
2128 { |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
2129 levels(s) <- trim(levels(s)) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
2130 if(recode.factor) { |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
2131 dots <- list(x=s, ...) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
2132 if(is.null(dots$sort)) dots$sort <- sort |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
2133 s <- do.call(what=reorder.factor, args=dots) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
2134 } |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
2135 s |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
2136 } |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
2137 |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
2138 trim.list <- function(s, recode.factor=TRUE, ...) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
2139 lapply(s, trim, recode.factor=recode.factor, ...) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
2140 |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
2141 trim.data.frame <- function(s, recode.factor=TRUE, ...) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
2142 { |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
2143 s[] <- trim.list(s, recode.factor=recode.factor, ...) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
2144 s |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
2145 } |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
2146 ####################################### |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
2147 # Compute the expected for each sequence-germline pair by codon |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
2148 getExpectedIndividualByCodon <- function(matInput){ |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
2149 if( any(grep("multicore",search())) ){ |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
2150 facGL <- factor(matInput[,2]) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
2151 facLevels = levels(facGL) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
2152 LisGLs_MutabilityU = mclapply(1:length(facLevels), function(x){ |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
2153 computeMutabilities(facLevels[x]) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
2154 }) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
2155 facIndex = match(facGL,facLevels) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
2156 |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
2157 LisGLs_Mutability = mclapply(1:nrow(matInput), function(x){ |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
2158 cInput = rep(NA,nchar(matInput[x,1])) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
2159 cInput[s2c(matInput[x,1])!="N"] = 1 |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
2160 LisGLs_MutabilityU[[facIndex[x]]] * cInput |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
2161 }) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
2162 |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
2163 LisGLs_Targeting = mclapply(1:dim(matInput)[1], function(x){ |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
2164 computeTargeting(matInput[x,2],LisGLs_Mutability[[x]]) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
2165 }) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
2166 |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
2167 LisGLs_MutationTypes = mclapply(1:length(matInput[,2]),function(x){ |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
2168 #print(x) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
2169 computeMutationTypes(matInput[x,2]) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
2170 }) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
2171 |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
2172 LisGLs_R_Exp = mclapply(1:nrow(matInput), function(x){ |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
2173 Exp_R <- rollapply(as.zoo(1:readEnd),width=3,by=3, |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
2174 function(codonNucs){ |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
2175 RPos = which(LisGLs_MutationTypes[[x]][,codonNucs]=="R") |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
2176 sum( LisGLs_Targeting[[x]][,codonNucs][RPos], na.rm=T ) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
2177 } |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
2178 ) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
2179 }) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
2180 |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
2181 LisGLs_S_Exp = mclapply(1:nrow(matInput), function(x){ |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
2182 Exp_S <- rollapply(as.zoo(1:readEnd),width=3,by=3, |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
2183 function(codonNucs){ |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
2184 SPos = which(LisGLs_MutationTypes[[x]][,codonNucs]=="S") |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
2185 sum( LisGLs_Targeting[[x]][,codonNucs][SPos], na.rm=T ) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
2186 } |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
2187 ) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
2188 }) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
2189 |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
2190 Exp_R = matrix(unlist(LisGLs_R_Exp),nrow=nrow(matInput),ncol=readEnd/3,T) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
2191 Exp_S = matrix(unlist(LisGLs_S_Exp),nrow=nrow(matInput),ncol=readEnd/3,T) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
2192 return( list( "Expected_R"=Exp_R, "Expected_S"=Exp_S) ) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
2193 }else{ |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
2194 facGL <- factor(matInput[,2]) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
2195 facLevels = levels(facGL) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
2196 LisGLs_MutabilityU = lapply(1:length(facLevels), function(x){ |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
2197 computeMutabilities(facLevels[x]) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
2198 }) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
2199 facIndex = match(facGL,facLevels) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
2200 |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
2201 LisGLs_Mutability = lapply(1:nrow(matInput), function(x){ |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
2202 cInput = rep(NA,nchar(matInput[x,1])) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
2203 cInput[s2c(matInput[x,1])!="N"] = 1 |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
2204 LisGLs_MutabilityU[[facIndex[x]]] * cInput |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
2205 }) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
2206 |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
2207 LisGLs_Targeting = lapply(1:dim(matInput)[1], function(x){ |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
2208 computeTargeting(matInput[x,2],LisGLs_Mutability[[x]]) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
2209 }) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
2210 |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
2211 LisGLs_MutationTypes = lapply(1:length(matInput[,2]),function(x){ |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
2212 #print(x) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
2213 computeMutationTypes(matInput[x,2]) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
2214 }) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
2215 |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
2216 LisGLs_R_Exp = lapply(1:nrow(matInput), function(x){ |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
2217 Exp_R <- rollapply(as.zoo(1:readEnd),width=3,by=3, |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
2218 function(codonNucs){ |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
2219 RPos = which(LisGLs_MutationTypes[[x]][,codonNucs]=="R") |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
2220 sum( LisGLs_Targeting[[x]][,codonNucs][RPos], na.rm=T ) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
2221 } |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
2222 ) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
2223 }) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
2224 |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
2225 LisGLs_S_Exp = lapply(1:nrow(matInput), function(x){ |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
2226 Exp_S <- rollapply(as.zoo(1:readEnd),width=3,by=3, |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
2227 function(codonNucs){ |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
2228 SPos = which(LisGLs_MutationTypes[[x]][,codonNucs]=="S") |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
2229 sum( LisGLs_Targeting[[x]][,codonNucs][SPos], na.rm=T ) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
2230 } |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
2231 ) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
2232 }) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
2233 |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
2234 Exp_R = matrix(unlist(LisGLs_R_Exp),nrow=nrow(matInput),ncol=readEnd/3,T) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
2235 Exp_S = matrix(unlist(LisGLs_S_Exp),nrow=nrow(matInput),ncol=readEnd/3,T) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
2236 return( list( "Expected_R"=Exp_R, "Expected_S"=Exp_S) ) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
2237 } |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
2238 } |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
2239 |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
2240 # getObservedMutationsByCodon <- function(listMutations){ |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
2241 # numbSeqs <- length(listMutations) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
2242 # obsMu_R <- matrix(0,nrow=numbSeqs,ncol=readEnd/3,dimnames=list(c(1:numbSeqs),c(1:(readEnd/3)))) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
2243 # obsMu_S <- obsMu_R |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
2244 # temp <- mclapply(1:length(listMutations), function(i){ |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
2245 # arrMutations = listMutations[[i]] |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
2246 # RPos = as.numeric(names(arrMutations)[arrMutations=="R"]) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
2247 # RPos <- sapply(RPos,getCodonNumb) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
2248 # if(any(RPos)){ |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
2249 # tabR <- table(RPos) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
2250 # obsMu_R[i,as.numeric(names(tabR))] <<- tabR |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
2251 # } |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
2252 # |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
2253 # SPos = as.numeric(names(arrMutations)[arrMutations=="S"]) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
2254 # SPos <- sapply(SPos,getCodonNumb) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
2255 # if(any(SPos)){ |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
2256 # tabS <- table(SPos) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
2257 # obsMu_S[i,names(tabS)] <<- tabS |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
2258 # } |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
2259 # } |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
2260 # ) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
2261 # return( list( "Observed_R"=obsMu_R, "Observed_S"=obsMu_S) ) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
2262 # } |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
2263 |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
2264 getObservedMutationsByCodon <- function(listMutations){ |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
2265 numbSeqs <- length(listMutations) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
2266 obsMu_R <- matrix(0,nrow=numbSeqs,ncol=readEnd/3,dimnames=list(c(1:numbSeqs),c(1:(readEnd/3)))) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
2267 obsMu_S <- obsMu_R |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
2268 temp <- lapply(1:length(listMutations), function(i){ |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
2269 arrMutations = listMutations[[i]] |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
2270 RPos = as.numeric(names(arrMutations)[arrMutations=="R"]) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
2271 RPos <- sapply(RPos,getCodonNumb) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
2272 if(any(RPos)){ |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
2273 tabR <- table(RPos) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
2274 obsMu_R[i,as.numeric(names(tabR))] <<- tabR |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
2275 } |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
2276 |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
2277 SPos = as.numeric(names(arrMutations)[arrMutations=="S"]) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
2278 SPos <- sapply(SPos,getCodonNumb) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
2279 if(any(SPos)){ |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
2280 tabS <- table(SPos) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
2281 obsMu_S[i,names(tabS)] <<- tabS |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
2282 } |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
2283 } |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
2284 ) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
2285 return( list( "Observed_R"=obsMu_R, "Observed_S"=obsMu_S) ) |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
2286 } |
|
64d74ba01a7c
"planemo upload commit 78d1fae87dbcf490e49a9f99e7a06de7328e16d4"
rhpvorderman
parents:
diff
changeset
|
2287 |
