Mercurial > repos > recetox > waveica

annotate waveica_wrapper.R @ 1:2bcfd5b450bb draft

Find changesets by keywords (author, files, the commit message), revision number or hash, or revset expression.
"planemo upload for repository https://github.com/RECETOX/galaxytools/tree/master/tools/waveica commit c52cdb1ab3c4e65ce3b55e29739f16abb8bfd6eb"
author recetox
date Wed, 28 Jul 2021 11:58:20 +0000
parents 2461d20911c9
children d08deef1eb44
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
rev   line source
0
2461d20911c9 "planemo upload for repository https://github.com/RECETOX/galaxytools/tree/master/tools/waveica commit 91376ea7a6736351b0cc086ca1bc6c553fdcda97"
recetox
parents:
diff changeset
1 waveica <- function(
2461d20911c9 "planemo upload for repository https://github.com/RECETOX/galaxytools/tree/master/tools/waveica commit 91376ea7a6736351b0cc086ca1bc6c553fdcda97"
recetox
parents:
diff changeset
2 data,
2461d20911c9 "planemo upload for repository https://github.com/RECETOX/galaxytools/tree/master/tools/waveica commit 91376ea7a6736351b0cc086ca1bc6c553fdcda97"
recetox
parents:
diff changeset
3 wavelet_filter,
2461d20911c9 "planemo upload for repository https://github.com/RECETOX/galaxytools/tree/master/tools/waveica commit 91376ea7a6736351b0cc086ca1bc6c553fdcda97"
recetox
parents:
diff changeset
4 wavelet_length,
2461d20911c9 "planemo upload for repository https://github.com/RECETOX/galaxytools/tree/master/tools/waveica commit 91376ea7a6736351b0cc086ca1bc6c553fdcda97"
recetox
parents:
diff changeset
5 k,
2461d20911c9 "planemo upload for repository https://github.com/RECETOX/galaxytools/tree/master/tools/waveica commit 91376ea7a6736351b0cc086ca1bc6c553fdcda97"
recetox
parents:
diff changeset
6 t,
2461d20911c9 "planemo upload for repository https://github.com/RECETOX/galaxytools/tree/master/tools/waveica commit 91376ea7a6736351b0cc086ca1bc6c553fdcda97"
recetox
parents:
diff changeset
7 t2,
2461d20911c9 "planemo upload for repository https://github.com/RECETOX/galaxytools/tree/master/tools/waveica commit 91376ea7a6736351b0cc086ca1bc6c553fdcda97"
recetox
parents:
diff changeset
8 alpha,
2461d20911c9 "planemo upload for repository https://github.com/RECETOX/galaxytools/tree/master/tools/waveica commit 91376ea7a6736351b0cc086ca1bc6c553fdcda97"
recetox
parents:
diff changeset
9 exclude_blanks
2461d20911c9 "planemo upload for repository https://github.com/RECETOX/galaxytools/tree/master/tools/waveica commit 91376ea7a6736351b0cc086ca1bc6c553fdcda97"
recetox
parents:
diff changeset
10 ) {
2461d20911c9 "planemo upload for repository https://github.com/RECETOX/galaxytools/tree/master/tools/waveica commit 91376ea7a6736351b0cc086ca1bc6c553fdcda97"
recetox
parents:
diff changeset
11
2461d20911c9 "planemo upload for repository https://github.com/RECETOX/galaxytools/tree/master/tools/waveica commit 91376ea7a6736351b0cc086ca1bc6c553fdcda97"
recetox
parents:
diff changeset
12 # get input from the Galaxy, preprocess data
1
2bcfd5b450bb "planemo upload for repository https://github.com/RECETOX/galaxytools/tree/master/tools/waveica commit c52cdb1ab3c4e65ce3b55e29739f16abb8bfd6eb"
recetox
parents: 0
diff changeset
13 data <- read.csv(data, header = TRUE, row.names = "sampleName")
2bcfd5b450bb "planemo upload for repository https://github.com/RECETOX/galaxytools/tree/master/tools/waveica commit c52cdb1ab3c4e65ce3b55e29739f16abb8bfd6eb"
recetox
parents: 0
diff changeset
14
2bcfd5b450bb "planemo upload for repository https://github.com/RECETOX/galaxytools/tree/master/tools/waveica commit c52cdb1ab3c4e65ce3b55e29739f16abb8bfd6eb"
recetox
parents: 0
diff changeset
15 # sort data by injection order
2bcfd5b450bb "planemo upload for repository https://github.com/RECETOX/galaxytools/tree/master/tools/waveica commit c52cdb1ab3c4e65ce3b55e29739f16abb8bfd6eb"
recetox
parents: 0
diff changeset
16 data <- data[order(data$injectionOrder, decreasing = FALSE), ]
2bcfd5b450bb "planemo upload for repository https://github.com/RECETOX/galaxytools/tree/master/tools/waveica commit c52cdb1ab3c4e65ce3b55e29739f16abb8bfd6eb"
recetox
parents: 0
diff changeset
17
2bcfd5b450bb "planemo upload for repository https://github.com/RECETOX/galaxytools/tree/master/tools/waveica commit c52cdb1ab3c4e65ce3b55e29739f16abb8bfd6eb"
recetox
parents: 0
diff changeset
18 data <- enumerate_groups(data)
0
2461d20911c9 "planemo upload for repository https://github.com/RECETOX/galaxytools/tree/master/tools/waveica commit 91376ea7a6736351b0cc086ca1bc6c553fdcda97"
recetox
parents:
diff changeset
19
2461d20911c9 "planemo upload for repository https://github.com/RECETOX/galaxytools/tree/master/tools/waveica commit 91376ea7a6736351b0cc086ca1bc6c553fdcda97"
recetox
parents:
diff changeset
20 # remove blanks from dataset
2461d20911c9 "planemo upload for repository https://github.com/RECETOX/galaxytools/tree/master/tools/waveica commit 91376ea7a6736351b0cc086ca1bc6c553fdcda97"
recetox
parents:
diff changeset
21 if (exclude_blanks) {
2461d20911c9 "planemo upload for repository https://github.com/RECETOX/galaxytools/tree/master/tools/waveica commit 91376ea7a6736351b0cc086ca1bc6c553fdcda97"
recetox
parents:
diff changeset
22 data <- exclude_group(data)
2461d20911c9 "planemo upload for repository https://github.com/RECETOX/galaxytools/tree/master/tools/waveica commit 91376ea7a6736351b0cc086ca1bc6c553fdcda97"
recetox
parents:
diff changeset
23 }
2461d20911c9 "planemo upload for repository https://github.com/RECETOX/galaxytools/tree/master/tools/waveica commit 91376ea7a6736351b0cc086ca1bc6c553fdcda97"
recetox
parents:
diff changeset
24
2461d20911c9 "planemo upload for repository https://github.com/RECETOX/galaxytools/tree/master/tools/waveica commit 91376ea7a6736351b0cc086ca1bc6c553fdcda97"
recetox
parents:
diff changeset
25 # separate data into features, batch and group
2461d20911c9 "planemo upload for repository https://github.com/RECETOX/galaxytools/tree/master/tools/waveica commit 91376ea7a6736351b0cc086ca1bc6c553fdcda97"
recetox
parents:
diff changeset
26 features <- data[, -c(1:4)]
2461d20911c9 "planemo upload for repository https://github.com/RECETOX/galaxytools/tree/master/tools/waveica commit 91376ea7a6736351b0cc086ca1bc6c553fdcda97"
recetox
parents:
diff changeset
27 group <- as.numeric(data$class)
2461d20911c9 "planemo upload for repository https://github.com/RECETOX/galaxytools/tree/master/tools/waveica commit 91376ea7a6736351b0cc086ca1bc6c553fdcda97"
recetox
parents:
diff changeset
28 batch <- data$batch
2461d20911c9 "planemo upload for repository https://github.com/RECETOX/galaxytools/tree/master/tools/waveica commit 91376ea7a6736351b0cc086ca1bc6c553fdcda97"
recetox
parents:
diff changeset
29
2461d20911c9 "planemo upload for repository https://github.com/RECETOX/galaxytools/tree/master/tools/waveica commit 91376ea7a6736351b0cc086ca1bc6c553fdcda97"
recetox
parents:
diff changeset
30 # run WaveICA
2461d20911c9 "planemo upload for repository https://github.com/RECETOX/galaxytools/tree/master/tools/waveica commit 91376ea7a6736351b0cc086ca1bc6c553fdcda97"
recetox
parents:
diff changeset
31 normalized_data <- WaveICA::WaveICA(
2461d20911c9 "planemo upload for repository https://github.com/RECETOX/galaxytools/tree/master/tools/waveica commit 91376ea7a6736351b0cc086ca1bc6c553fdcda97"
recetox
parents:
diff changeset
32 data = features,
2461d20911c9 "planemo upload for repository https://github.com/RECETOX/galaxytools/tree/master/tools/waveica commit 91376ea7a6736351b0cc086ca1bc6c553fdcda97"
recetox
parents:
diff changeset
33 wf = get_wf(wavelet_filter, wavelet_length),
2461d20911c9 "planemo upload for repository https://github.com/RECETOX/galaxytools/tree/master/tools/waveica commit 91376ea7a6736351b0cc086ca1bc6c553fdcda97"
recetox
parents:
diff changeset
34 batch = batch,
2461d20911c9 "planemo upload for repository https://github.com/RECETOX/galaxytools/tree/master/tools/waveica commit 91376ea7a6736351b0cc086ca1bc6c553fdcda97"
recetox
parents:
diff changeset
35 group = group,
2461d20911c9 "planemo upload for repository https://github.com/RECETOX/galaxytools/tree/master/tools/waveica commit 91376ea7a6736351b0cc086ca1bc6c553fdcda97"
recetox
parents:
diff changeset
36 K = k,
2461d20911c9 "planemo upload for repository https://github.com/RECETOX/galaxytools/tree/master/tools/waveica commit 91376ea7a6736351b0cc086ca1bc6c553fdcda97"
recetox
parents:
diff changeset
37 t = t,
2461d20911c9 "planemo upload for repository https://github.com/RECETOX/galaxytools/tree/master/tools/waveica commit 91376ea7a6736351b0cc086ca1bc6c553fdcda97"
recetox
parents:
diff changeset
38 t2 = t2,
2461d20911c9 "planemo upload for repository https://github.com/RECETOX/galaxytools/tree/master/tools/waveica commit 91376ea7a6736351b0cc086ca1bc6c553fdcda97"
recetox
parents:
diff changeset
39 alpha = alpha
2461d20911c9 "planemo upload for repository https://github.com/RECETOX/galaxytools/tree/master/tools/waveica commit 91376ea7a6736351b0cc086ca1bc6c553fdcda97"
recetox
parents:
diff changeset
40 )
2461d20911c9 "planemo upload for repository https://github.com/RECETOX/galaxytools/tree/master/tools/waveica commit 91376ea7a6736351b0cc086ca1bc6c553fdcda97"
recetox
parents:
diff changeset
41
2461d20911c9 "planemo upload for repository https://github.com/RECETOX/galaxytools/tree/master/tools/waveica commit 91376ea7a6736351b0cc086ca1bc6c553fdcda97"
recetox
parents:
diff changeset
42 return(normalized_data)
2461d20911c9 "planemo upload for repository https://github.com/RECETOX/galaxytools/tree/master/tools/waveica commit 91376ea7a6736351b0cc086ca1bc6c553fdcda97"
recetox
parents:
diff changeset
43 }
2461d20911c9 "planemo upload for repository https://github.com/RECETOX/galaxytools/tree/master/tools/waveica commit 91376ea7a6736351b0cc086ca1bc6c553fdcda97"
recetox
parents:
diff changeset
44
2461d20911c9 "planemo upload for repository https://github.com/RECETOX/galaxytools/tree/master/tools/waveica commit 91376ea7a6736351b0cc086ca1bc6c553fdcda97"
recetox
parents:
diff changeset
45
1
2bcfd5b450bb "planemo upload for repository https://github.com/RECETOX/galaxytools/tree/master/tools/waveica commit c52cdb1ab3c4e65ce3b55e29739f16abb8bfd6eb"
recetox
parents: 0
diff changeset
46 # Match group labels with [blank/sample/qc] and enumerate them
2bcfd5b450bb "planemo upload for repository https://github.com/RECETOX/galaxytools/tree/master/tools/waveica commit c52cdb1ab3c4e65ce3b55e29739f16abb8bfd6eb"
recetox
parents: 0
diff changeset
47 enumerate_groups <- function(data) {
0
2461d20911c9 "planemo upload for repository https://github.com/RECETOX/galaxytools/tree/master/tools/waveica commit 91376ea7a6736351b0cc086ca1bc6c553fdcda97"
recetox
parents:
diff changeset
48
1
2bcfd5b450bb "planemo upload for repository https://github.com/RECETOX/galaxytools/tree/master/tools/waveica commit c52cdb1ab3c4e65ce3b55e29739f16abb8bfd6eb"
recetox
parents: 0
diff changeset
49 data$sampleType[grepl("blank", tolower(data$sampleType))] <- 0
2bcfd5b450bb "planemo upload for repository https://github.com/RECETOX/galaxytools/tree/master/tools/waveica commit c52cdb1ab3c4e65ce3b55e29739f16abb8bfd6eb"
recetox
parents: 0
diff changeset
50 data$sampleType[grepl("sample", tolower(data$sampleType))] <- 1
2bcfd5b450bb "planemo upload for repository https://github.com/RECETOX/galaxytools/tree/master/tools/waveica commit c52cdb1ab3c4e65ce3b55e29739f16abb8bfd6eb"
recetox
parents: 0
diff changeset
51 data$sampleType[grepl("qc", tolower(data$sampleType))] <- 2
0
2461d20911c9 "planemo upload for repository https://github.com/RECETOX/galaxytools/tree/master/tools/waveica commit 91376ea7a6736351b0cc086ca1bc6c553fdcda97"
recetox
parents:
diff changeset
52
2461d20911c9 "planemo upload for repository https://github.com/RECETOX/galaxytools/tree/master/tools/waveica commit 91376ea7a6736351b0cc086ca1bc6c553fdcda97"
recetox
parents:
diff changeset
53 return(data)
2461d20911c9 "planemo upload for repository https://github.com/RECETOX/galaxytools/tree/master/tools/waveica commit 91376ea7a6736351b0cc086ca1bc6c553fdcda97"
recetox
parents:
diff changeset
54 }
2461d20911c9 "planemo upload for repository https://github.com/RECETOX/galaxytools/tree/master/tools/waveica commit 91376ea7a6736351b0cc086ca1bc6c553fdcda97"
recetox
parents:
diff changeset
55
2461d20911c9 "planemo upload for repository https://github.com/RECETOX/galaxytools/tree/master/tools/waveica commit 91376ea7a6736351b0cc086ca1bc6c553fdcda97"
recetox
parents:
diff changeset
56
2461d20911c9 "planemo upload for repository https://github.com/RECETOX/galaxytools/tree/master/tools/waveica commit 91376ea7a6736351b0cc086ca1bc6c553fdcda97"
recetox
parents:
diff changeset
57 # Create appropriate input for R wavelets function
2461d20911c9 "planemo upload for repository https://github.com/RECETOX/galaxytools/tree/master/tools/waveica commit 91376ea7a6736351b0cc086ca1bc6c553fdcda97"
recetox
parents:
diff changeset
58 get_wf <- function(wavelet_filter, wavelet_length) {
2461d20911c9 "planemo upload for repository https://github.com/RECETOX/galaxytools/tree/master/tools/waveica commit 91376ea7a6736351b0cc086ca1bc6c553fdcda97"
recetox
parents:
diff changeset
59 wf <- paste(wavelet_filter, wavelet_length, sep = "")
2461d20911c9 "planemo upload for repository https://github.com/RECETOX/galaxytools/tree/master/tools/waveica commit 91376ea7a6736351b0cc086ca1bc6c553fdcda97"
recetox
parents:
diff changeset
60
2461d20911c9 "planemo upload for repository https://github.com/RECETOX/galaxytools/tree/master/tools/waveica commit 91376ea7a6736351b0cc086ca1bc6c553fdcda97"
recetox
parents:
diff changeset
61 # exception to the wavelet function
2461d20911c9 "planemo upload for repository https://github.com/RECETOX/galaxytools/tree/master/tools/waveica commit 91376ea7a6736351b0cc086ca1bc6c553fdcda97"
recetox
parents:
diff changeset
62 if (wf == "d2") {
2461d20911c9 "planemo upload for repository https://github.com/RECETOX/galaxytools/tree/master/tools/waveica commit 91376ea7a6736351b0cc086ca1bc6c553fdcda97"
recetox
parents:
diff changeset
63 wf <- "haar"
2461d20911c9 "planemo upload for repository https://github.com/RECETOX/galaxytools/tree/master/tools/waveica commit 91376ea7a6736351b0cc086ca1bc6c553fdcda97"
recetox
parents:
diff changeset
64 }
2461d20911c9 "planemo upload for repository https://github.com/RECETOX/galaxytools/tree/master/tools/waveica commit 91376ea7a6736351b0cc086ca1bc6c553fdcda97"
recetox
parents:
diff changeset
65
2461d20911c9 "planemo upload for repository https://github.com/RECETOX/galaxytools/tree/master/tools/waveica commit 91376ea7a6736351b0cc086ca1bc6c553fdcda97"
recetox
parents:
diff changeset
66 return(wf)
2461d20911c9 "planemo upload for repository https://github.com/RECETOX/galaxytools/tree/master/tools/waveica commit 91376ea7a6736351b0cc086ca1bc6c553fdcda97"
recetox
parents:
diff changeset
67 }
2461d20911c9 "planemo upload for repository https://github.com/RECETOX/galaxytools/tree/master/tools/waveica commit 91376ea7a6736351b0cc086ca1bc6c553fdcda97"
recetox
parents:
diff changeset
68
2461d20911c9 "planemo upload for repository https://github.com/RECETOX/galaxytools/tree/master/tools/waveica commit 91376ea7a6736351b0cc086ca1bc6c553fdcda97"
recetox
parents:
diff changeset
69
2461d20911c9 "planemo upload for repository https://github.com/RECETOX/galaxytools/tree/master/tools/waveica commit 91376ea7a6736351b0cc086ca1bc6c553fdcda97"
recetox
parents:
diff changeset
70 # Exclude blanks from a dataframe
2461d20911c9 "planemo upload for repository https://github.com/RECETOX/galaxytools/tree/master/tools/waveica commit 91376ea7a6736351b0cc086ca1bc6c553fdcda97"
recetox
parents:
diff changeset
71 exclude_group <- function(data) {
2461d20911c9 "planemo upload for repository https://github.com/RECETOX/galaxytools/tree/master/tools/waveica commit 91376ea7a6736351b0cc086ca1bc6c553fdcda97"
recetox
parents:
diff changeset
72 row_idx_to_exclude <- which(data$class %in% 0)
1
2bcfd5b450bb "planemo upload for repository https://github.com/RECETOX/galaxytools/tree/master/tools/waveica commit c52cdb1ab3c4e65ce3b55e29739f16abb8bfd6eb"
recetox
parents: 0
diff changeset
73 if (length(row_idx_to_exclude) > 0) {
0
2461d20911c9 "planemo upload for repository https://github.com/RECETOX/galaxytools/tree/master/tools/waveica commit 91376ea7a6736351b0cc086ca1bc6c553fdcda97"
recetox
parents:
diff changeset
74 data_without_blanks <- data[-c(row_idx_to_exclude), ]
2461d20911c9 "planemo upload for repository https://github.com/RECETOX/galaxytools/tree/master/tools/waveica commit 91376ea7a6736351b0cc086ca1bc6c553fdcda97"
recetox
parents:
diff changeset
75 msg <- paste("Blank samples have been excluded from the dataframe.\n")
2461d20911c9 "planemo upload for repository https://github.com/RECETOX/galaxytools/tree/master/tools/waveica commit 91376ea7a6736351b0cc086ca1bc6c553fdcda97"
recetox
parents:
diff changeset
76 cat(msg)
2461d20911c9 "planemo upload for repository https://github.com/RECETOX/galaxytools/tree/master/tools/waveica commit 91376ea7a6736351b0cc086ca1bc6c553fdcda97"
recetox
parents:
diff changeset
77 return(data_without_blanks)
2461d20911c9 "planemo upload for repository https://github.com/RECETOX/galaxytools/tree/master/tools/waveica commit 91376ea7a6736351b0cc086ca1bc6c553fdcda97"
recetox
parents:
diff changeset
78 }
2461d20911c9 "planemo upload for repository https://github.com/RECETOX/galaxytools/tree/master/tools/waveica commit 91376ea7a6736351b0cc086ca1bc6c553fdcda97"
recetox
parents:
diff changeset
79 else {
2461d20911c9 "planemo upload for repository https://github.com/RECETOX/galaxytools/tree/master/tools/waveica commit 91376ea7a6736351b0cc086ca1bc6c553fdcda97"
recetox
parents:
diff changeset
80 return(data)
2461d20911c9 "planemo upload for repository https://github.com/RECETOX/galaxytools/tree/master/tools/waveica commit 91376ea7a6736351b0cc086ca1bc6c553fdcda97"
recetox
parents:
diff changeset
81 }
2461d20911c9 "planemo upload for repository https://github.com/RECETOX/galaxytools/tree/master/tools/waveica commit 91376ea7a6736351b0cc086ca1bc6c553fdcda97"
recetox
parents:
diff changeset
82 }
2461d20911c9 "planemo upload for repository https://github.com/RECETOX/galaxytools/tree/master/tools/waveica commit 91376ea7a6736351b0cc086ca1bc6c553fdcda97"
recetox
parents:
diff changeset
83
2461d20911c9 "planemo upload for repository https://github.com/RECETOX/galaxytools/tree/master/tools/waveica commit 91376ea7a6736351b0cc086ca1bc6c553fdcda97"
recetox
parents:
diff changeset
84
2461d20911c9 "planemo upload for repository https://github.com/RECETOX/galaxytools/tree/master/tools/waveica commit 91376ea7a6736351b0cc086ca1bc6c553fdcda97"
recetox
parents:
diff changeset
85 # Store output of WaveICA in a tsv file
2461d20911c9 "planemo upload for repository https://github.com/RECETOX/galaxytools/tree/master/tools/waveica commit 91376ea7a6736351b0cc086ca1bc6c553fdcda97"
recetox
parents:
diff changeset
86 store_data <- function(normalized_data, output) {
2461d20911c9 "planemo upload for repository https://github.com/RECETOX/galaxytools/tree/master/tools/waveica commit 91376ea7a6736351b0cc086ca1bc6c553fdcda97"
recetox
parents:
diff changeset
87 write.table(normalized_data, file = output, sep = "\t", col.names = NA)
2461d20911c9 "planemo upload for repository https://github.com/RECETOX/galaxytools/tree/master/tools/waveica commit 91376ea7a6736351b0cc086ca1bc6c553fdcda97"
recetox
parents:
diff changeset
88 cat("Normalization has been completed.\n")
2461d20911c9 "planemo upload for repository https://github.com/RECETOX/galaxytools/tree/master/tools/waveica commit 91376ea7a6736351b0cc086ca1bc6c553fdcda97"
recetox
parents:
diff changeset
89 }