Mercurial > repos > abims-sbr > cds_search

annotate scripts/S01_find_orf_on_multiple_alignment.py @ 9:640ef4c06ed5 draft

Find changesets by keywords (author, files, the commit message), revision number or hash, or revset expression.
planemo upload for repository https://github.com/abims-sbr/adaptsearch commit f1ba8d136e0129f3e8435b25a95f70f697d51464-dirty
author abims-sbr
date Tue, 03 Jul 2018 10:54:18 -0400
parents 716a45028e55
children 3d00be2d05f3
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
rev   line source
1
567d5b771a90 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
1 #!/usr/bin/python
8
716a45028e55 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit 90cfcf697b9f128e81bea1270378e59d63ab0a6f
abims-sbr
parents: 7
diff changeset
2 # coding: utf8
1
567d5b771a90 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
3 ## Author: Eric Fontanillas
8
716a45028e55 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit 90cfcf697b9f128e81bea1270378e59d63ab0a6f
abims-sbr
parents: 7
diff changeset
4 ## Modification: 03/09/14 by Julie BAFFARD
716a45028e55 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit 90cfcf697b9f128e81bea1270378e59d63ab0a6f
abims-sbr
parents: 7
diff changeset
5 ## Last modification : 05/03/18 by Victor Mataigne
1
567d5b771a90 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
6
567d5b771a90 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
7 ## Description: Predict potential ORF on the basis of 2 criteria + 1 optional criteria
567d5b771a90 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
8 ## CRITERIA 1 ## Longest part of the alignment of sequence without codon stop "*", tested in the 3 potential ORF
567d5b771a90 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
9 ## CRITERIA 2 ## This longest part should be > 150nc or 50aa
567d5b771a90 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
10 ## CRITERIA 3 ## [OPTIONNAL] A codon start "M" should be present in this longuest part, before the last 50 aa
567d5b771a90 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
11 ## OUTPUTs "05_CDS_aa" & "05_CDS_nuc" => NOT INCLUDE THIS CRITERIA
567d5b771a90 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
12 ## OUTPUTs "06_CDS_with_M_aa" & "06_CDS_with_M_nuc" => INCLUDE THIS CRITERIA
567d5b771a90 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
13
567d5b771a90 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
14 ####################################################
567d5b771a90 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
15 ###### DEF 2 : Create bash for genetic code ########
567d5b771a90 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
16 ####################################################
567d5b771a90 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
17 ### KEY = codon
567d5b771a90 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
18 ### VALUE = Amino Acid
567d5b771a90 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
19
567d5b771a90 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
20 def code_universel(F1):
567d5b771a90 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
21 bash_codeUniversel = {}
7
35e39b4128ba planemo upload for repository https://github.com/abims-sbr/adaptsearch commit b7a3030ea134b5dfad89b1a869db659d72d1145c
abims-sbr
parents: 3
diff changeset
22 with open(F1, "r") as file:
35e39b4128ba planemo upload for repository https://github.com/abims-sbr/adaptsearch commit b7a3030ea134b5dfad89b1a869db659d72d1145c
abims-sbr
parents: 3
diff changeset
23 for line in file.readlines():
35e39b4128ba planemo upload for repository https://github.com/abims-sbr/adaptsearch commit b7a3030ea134b5dfad89b1a869db659d72d1145c
abims-sbr
parents: 3
diff changeset
24 L1 = string.split(line, " ")
35e39b4128ba planemo upload for repository https://github.com/abims-sbr/adaptsearch commit b7a3030ea134b5dfad89b1a869db659d72d1145c
abims-sbr
parents: 3
diff changeset
25 length1 = len(L1)
35e39b4128ba planemo upload for repository https://github.com/abims-sbr/adaptsearch commit b7a3030ea134b5dfad89b1a869db659d72d1145c
abims-sbr
parents: 3
diff changeset
26 if length1 == 3:
35e39b4128ba planemo upload for repository https://github.com/abims-sbr/adaptsearch commit b7a3030ea134b5dfad89b1a869db659d72d1145c
abims-sbr
parents: 3
diff changeset
27 key = L1[0]
35e39b4128ba planemo upload for repository https://github.com/abims-sbr/adaptsearch commit b7a3030ea134b5dfad89b1a869db659d72d1145c
abims-sbr
parents: 3
diff changeset
28 value = L1[2][:-1]
35e39b4128ba planemo upload for repository https://github.com/abims-sbr/adaptsearch commit b7a3030ea134b5dfad89b1a869db659d72d1145c
abims-sbr
parents: 3
diff changeset
29 bash_codeUniversel[key] = value
35e39b4128ba planemo upload for repository https://github.com/abims-sbr/adaptsearch commit b7a3030ea134b5dfad89b1a869db659d72d1145c
abims-sbr
parents: 3
diff changeset
30 else:
35e39b4128ba planemo upload for repository https://github.com/abims-sbr/adaptsearch commit b7a3030ea134b5dfad89b1a869db659d72d1145c
abims-sbr
parents: 3
diff changeset
31 key = L1[0]
35e39b4128ba planemo upload for repository https://github.com/abims-sbr/adaptsearch commit b7a3030ea134b5dfad89b1a869db659d72d1145c
abims-sbr
parents: 3
diff changeset
32 value = L1[2]
35e39b4128ba planemo upload for repository https://github.com/abims-sbr/adaptsearch commit b7a3030ea134b5dfad89b1a869db659d72d1145c
abims-sbr
parents: 3
diff changeset
33 bash_codeUniversel[key] = value
1
567d5b771a90 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
34 return(bash_codeUniversel)
567d5b771a90 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
35 ###########################################################
567d5b771a90 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
36
567d5b771a90 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
37
567d5b771a90 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
38 ######################################################################################################################
2
0d2f72caea10 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit 44a89d5eeb82789bfc643b33c11f391281b6374b
abims-sbr
parents: 1
diff changeset
39 ##### DEF 3 : Test if the sequence is a multiple of 3, and if not correct the sequence to become a multiple of 3 #####
1
567d5b771a90 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
40 ######################################################################################################################
567d5b771a90 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
41 ### WEAKNESS OF THAT APPROACH = I remove extra base(s) at the end of the sequence ==> I can lost a codon, when I test ORF (as I will decay the ORF)
567d5b771a90 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
42 def multiple3(seq):
567d5b771a90 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
43 leng = len(seq)
567d5b771a90 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
44 modulo = leng%3
567d5b771a90 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
45 if modulo == 0: # the results of dividing leng per 3 is an integer
2
0d2f72caea10 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit 44a89d5eeb82789bfc643b33c11f391281b6374b
abims-sbr
parents: 1
diff changeset
46 new_seq = seq
1
567d5b771a90 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
47 elif modulo == 1: # means 1 extra nc (nucleotid) needs to be removed (the remaining of modulo indicate the part which is non-dividable per 3)
567d5b771a90 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
48 new_seq = seq[:-1] # remove the last nc
567d5b771a90 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
49 elif modulo == 2: # means 2 extra nc (nucleotid) needs to be removed (the remaining of modulo indicate the part which is non-dividable per 3)
567d5b771a90 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
50 new_seq = seq[:-2] # remove the 2 last nc
2
0d2f72caea10 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit 44a89d5eeb82789bfc643b33c11f391281b6374b
abims-sbr
parents: 1
diff changeset
51 len1 = len(new_seq)
1
567d5b771a90 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
52 return(new_seq, modulo)
567d5b771a90 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
53 ##########################################################
567d5b771a90 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
54
567d5b771a90 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
55
567d5b771a90 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
56 #############################
567d5b771a90 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
57 ###### DEF 4 : GET ORF ######
567d5b771a90 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
58 #############################
567d5b771a90 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
59 ##- MULTIPLE SEQUENCE BASED : Based on ALIGNMENT of several sequences
567d5b771a90 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
60 ##- CRITERIA1: Get the segment in the alignment with no codon stop
567d5b771a90 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
61
567d5b771a90 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
62
567d5b771a90 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
63 ###### DEF 4 - Part 1 - ######
567d5b771a90 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
64 ##############################
567d5b771a90 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
65 def simply_get_ORF(seq_dna, bash_codeUniversel):
567d5b771a90 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
66 seq_aa = ""
567d5b771a90 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
67 i = 0
567d5b771a90 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
68 len1 = len(seq_dna)
567d5b771a90 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
69 while i < len1:
567d5b771a90 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
70 base1 = seq_dna[i]
567d5b771a90 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
71 base1 = string.capitalize(base1)
567d5b771a90 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
72 base2 = seq_dna[i+1]
567d5b771a90 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
73 base2 = string.capitalize(base2)
567d5b771a90 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
74 base3 = seq_dna[i+2]
567d5b771a90 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
75 base3 = string.capitalize(base3)
2
0d2f72caea10 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit 44a89d5eeb82789bfc643b33c11f391281b6374b
abims-sbr
parents: 1
diff changeset
76
1
567d5b771a90 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
77 codon = base1+base2+base3
567d5b771a90 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
78 codon = string.replace(codon, "T", "U")
567d5b771a90 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
79
567d5b771a90 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
80 if codon in bash_codeUniversel.keys():
567d5b771a90 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
81 aa = bash_codeUniversel[codon]
2
0d2f72caea10 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit 44a89d5eeb82789bfc643b33c11f391281b6374b
abims-sbr
parents: 1
diff changeset
82 seq_aa = seq_aa + aa
1
567d5b771a90 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
83 else:
567d5b771a90 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
84 seq_aa = seq_aa +"?" ### Take account for gap "-" and "N"
567d5b771a90 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
85 i = i + 3
567d5b771a90 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
86
567d5b771a90 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
87 return(seq_aa)
567d5b771a90 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
88 ##########################################################
567d5b771a90 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
89
567d5b771a90 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
90
567d5b771a90 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
91 ###### DEF 4 - Part 2 - ######
567d5b771a90 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
92 ##############################
2
0d2f72caea10 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit 44a89d5eeb82789bfc643b33c11f391281b6374b
abims-sbr
parents: 1
diff changeset
93 def find_good_ORF_criteria_3(bash_aligned_nc_seq, bash_codeUniversel):
0d2f72caea10 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit 44a89d5eeb82789bfc643b33c11f391281b6374b
abims-sbr
parents: 1
diff changeset
94
1
567d5b771a90 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
95 ## 1 ## Get the list of aligned aa seq for the 3 ORF:
567d5b771a90 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
96 bash_of_aligned_aa_seq_3ORF = {}
567d5b771a90 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
97 bash_of_aligned_nuc_seq_3ORF = {}
567d5b771a90 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
98 BEST_LONGUEST_SUBSEQUENCE_LIST_POSITION = []
567d5b771a90 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
99 for fasta_name in bash_aligned_nc_seq.keys():
567d5b771a90 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
100 ## 1.1. ## Get the raw sequence
567d5b771a90 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
101 sequence_nc = bash_aligned_nc_seq[fasta_name]
567d5b771a90 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
102
567d5b771a90 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
103 ## 1.2. ## Check whether the sequence is multiple of 3, and correct it if not:
567d5b771a90 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
104 new_sequence_nc, modulo = multiple3(sequence_nc) ### DEF 3 ###
2
0d2f72caea10 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit 44a89d5eeb82789bfc643b33c11f391281b6374b
abims-sbr
parents: 1
diff changeset
105
1
567d5b771a90 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
106 ## 1.3. ## Get the 3 ORFs (nuc) for each sequence
567d5b771a90 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
107 seq_nuc_ORF1 = new_sequence_nc
567d5b771a90 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
108 seq_nuc_ORF2 = new_sequence_nc[1:-2]
567d5b771a90 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
109 seq_nuc_ORF3 = new_sequence_nc[2:-1]
567d5b771a90 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
110 seq_reversed=ReverseComplement2(seq_nuc_ORF1)
567d5b771a90 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
111 seq_nuc_ORF4=seq_reversed
567d5b771a90 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
112 seq_nuc_ORF5=seq_reversed[1:-2]
567d5b771a90 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
113 seq_nuc_ORF6=seq_reversed[2:-1]
2
0d2f72caea10 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit 44a89d5eeb82789bfc643b33c11f391281b6374b
abims-sbr
parents: 1
diff changeset
114
1
567d5b771a90 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
115 LIST_6_ORF_nuc = [seq_nuc_ORF1, seq_nuc_ORF2, seq_nuc_ORF3,seq_nuc_ORF4,seq_nuc_ORF5,seq_nuc_ORF6]
567d5b771a90 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
116 bash_of_aligned_nuc_seq_3ORF[fasta_name] = LIST_6_ORF_nuc ### For each seq of the multialignment => give the 6 ORFs (in nuc)
567d5b771a90 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
117
567d5b771a90 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
118 ## 1.4. ## Get the 3 ORFs (aa) for each sequence
567d5b771a90 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
119 seq_prot_ORF1 = simply_get_ORF(seq_nuc_ORF1,bash_codeUniversel) ### DEF 4 - Part 1 - ##
567d5b771a90 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
120 seq_prot_ORF2 = simply_get_ORF(seq_nuc_ORF2,bash_codeUniversel) ### DEF 4 - Part 1 - ##
567d5b771a90 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
121 seq_prot_ORF3 = simply_get_ORF(seq_nuc_ORF3,bash_codeUniversel) ### DEF 4 - Part 1 - ##
2
0d2f72caea10 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit 44a89d5eeb82789bfc643b33c11f391281b6374b
abims-sbr
parents: 1
diff changeset
122 seq_prot_ORF4 = simply_get_ORF(seq_nuc_ORF4,bash_codeUniversel) ### DEF 4 - Part 1 - ##
1
567d5b771a90 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
123 seq_prot_ORF5 = simply_get_ORF(seq_nuc_ORF5,bash_codeUniversel) ### DEF 4 - Part 1 - ##
567d5b771a90 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
124 seq_prot_ORF6 = simply_get_ORF(seq_nuc_ORF6,bash_codeUniversel) ### DEF 4 - Part 1 - ##
567d5b771a90 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
125
567d5b771a90 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
126 LIST_6_ORF_aa = [seq_prot_ORF1, seq_prot_ORF2, seq_prot_ORF3,seq_prot_ORF4,seq_prot_ORF5,seq_prot_ORF6]
2
0d2f72caea10 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit 44a89d5eeb82789bfc643b33c11f391281b6374b
abims-sbr
parents: 1
diff changeset
127 bash_of_aligned_aa_seq_3ORF[fasta_name] = LIST_6_ORF_aa ### For each seq of the multialignment => give the 6 ORFs (in aa)
1
567d5b771a90 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
128
567d5b771a90 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
129 ## 2 ## Test for the best ORF (Get the longuest segment in the alignment with no codon stop ... for each ORF ... the longuest should give the ORF)
567d5b771a90 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
130 BEST_MAX = 0
2
0d2f72caea10 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit 44a89d5eeb82789bfc643b33c11f391281b6374b
abims-sbr
parents: 1
diff changeset
131 for i in [0,1,2,3,4,5]: ### Test the 6 ORFs
1
567d5b771a90 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
132 ORF_Aligned_aa = []
567d5b771a90 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
133 ORF_Aligned_nuc = []
2
0d2f72caea10 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit 44a89d5eeb82789bfc643b33c11f391281b6374b
abims-sbr
parents: 1
diff changeset
134
1
567d5b771a90 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
135
567d5b771a90 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
136 ## 2.1 ## Get the alignment of sequence for a given ORF
567d5b771a90 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
137 ## Compare the 1rst ORF between all sequence => list them in ORF_Aligned_aa // them do the same for the second ORF, and them the 3rd
2
0d2f72caea10 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit 44a89d5eeb82789bfc643b33c11f391281b6374b
abims-sbr
parents: 1
diff changeset
138 for fasta_name in bash_of_aligned_aa_seq_3ORF.keys():
1
567d5b771a90 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
139 ORFsequence = bash_of_aligned_aa_seq_3ORF[fasta_name][i]
567d5b771a90 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
140 aa_length = len(ORFsequence)
567d5b771a90 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
141 ORF_Aligned_aa.append(ORFsequence) ### List of all sequences in the ORF nb "i" =
567d5b771a90 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
142
567d5b771a90 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
143 n = i+1
2
0d2f72caea10 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit 44a89d5eeb82789bfc643b33c11f391281b6374b
abims-sbr
parents: 1
diff changeset
144
0d2f72caea10 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit 44a89d5eeb82789bfc643b33c11f391281b6374b
abims-sbr
parents: 1
diff changeset
145 for fasta_name in bash_of_aligned_nuc_seq_3ORF.keys():
1
567d5b771a90 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
146 ORFsequence = bash_of_aligned_nuc_seq_3ORF[fasta_name][i]
567d5b771a90 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
147 nuc_length = len(ORFsequence)
567d5b771a90 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
148 ORF_Aligned_nuc.append(ORFsequence) ### List of all sequences in the ORF nb "i" =
567d5b771a90 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
149
567d5b771a90 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
150 ## 2.2 ## Get the list of sublist of positions whithout codon stop in the alignment
567d5b771a90 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
151 ## For each ORF, now we have the list of sequences available (i.e. THE ALIGNMENT IN A GIVEN ORF)
567d5b771a90 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
152 ## Next step is to get the longuest subsequence whithout stop
567d5b771a90 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
153 ## We will explore the presence of stop "*" in each column of the alignment, and get the positions of the segments between the positions with "*"
567d5b771a90 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
154 MAX_LENGTH = 0
2
0d2f72caea10 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit 44a89d5eeb82789bfc643b33c11f391281b6374b
abims-sbr
parents: 1
diff changeset
155 LONGUEST_SEGMENT_UNSTOPPED = ""
0d2f72caea10 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit 44a89d5eeb82789bfc643b33c11f391281b6374b
abims-sbr
parents: 1
diff changeset
156 j = 0 # Start from first position in alignment
1
567d5b771a90 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
157 List_of_List_subsequences = []
567d5b771a90 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
158 List_positions_subsequence = []
2
0d2f72caea10 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit 44a89d5eeb82789bfc643b33c11f391281b6374b
abims-sbr
parents: 1
diff changeset
159 while j < aa_length:
1
567d5b771a90 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
160 column = []
567d5b771a90 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
161 for seq in ORF_Aligned_aa:
567d5b771a90 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
162 column.append(seq[j])
567d5b771a90 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
163 j = j+1
567d5b771a90 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
164 if "*" in column:
567d5b771a90 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
165 List_of_List_subsequences.append(List_positions_subsequence) ## Add previous list of positions
567d5b771a90 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
166 List_positions_subsequence = [] ## Re-initialyse list of positions
567d5b771a90 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
167 else:
567d5b771a90 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
168 List_positions_subsequence.append(j)
2
0d2f72caea10 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit 44a89d5eeb82789bfc643b33c11f391281b6374b
abims-sbr
parents: 1
diff changeset
169
1
567d5b771a90 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
170 ## 2.3 ## Among all the sublists (separated by column with codon stop "*"), get the longuest one (BETTER SEGMENT for a given ORF)
567d5b771a90 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
171 LONGUEST_SUBSEQUENCE_LIST_POSITION = []
567d5b771a90 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
172 MAX=0
567d5b771a90 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
173 for sublist in List_of_List_subsequences:
567d5b771a90 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
174 if len(sublist) > MAX and len(sublist) > MINIMAL_CDS_LENGTH:
567d5b771a90 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
175 MAX = len(sublist)
567d5b771a90 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
176 LONGUEST_SUBSEQUENCE_LIST_POSITION = sublist
2
0d2f72caea10 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit 44a89d5eeb82789bfc643b33c11f391281b6374b
abims-sbr
parents: 1
diff changeset
177
1
567d5b771a90 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
178 ## 2.4. ## Test if the longuest subsequence start exactly at the beginning of the original sequence (i.e. means the ORF maybe truncated)
567d5b771a90 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
179 if LONGUEST_SUBSEQUENCE_LIST_POSITION != []:
567d5b771a90 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
180 if LONGUEST_SUBSEQUENCE_LIST_POSITION[0] == 0:
567d5b771a90 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
181 CDS_maybe_truncated = 1
567d5b771a90 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
182 else:
567d5b771a90 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
183 CDS_maybe_truncated = 0
567d5b771a90 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
184 else:
567d5b771a90 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
185 CDS_maybe_truncated = 0
2
0d2f72caea10 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit 44a89d5eeb82789bfc643b33c11f391281b6374b
abims-sbr
parents: 1
diff changeset
186
1
567d5b771a90 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
187
567d5b771a90 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
188 ## 2.5 ## Test if this BETTER SEGMENT for a given ORF, is the better than the one for the other ORF (GET THE BEST ORF)
567d5b771a90 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
189 ## Test whether it is the better ORF
567d5b771a90 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
190 if MAX > BEST_MAX:
567d5b771a90 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
191 BEST_MAX = MAX
567d5b771a90 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
192 BEST_ORF = i+1
567d5b771a90 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
193 BEST_LONGUEST_SUBSEQUENCE_LIST_POSITION = LONGUEST_SUBSEQUENCE_LIST_POSITION
567d5b771a90 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
194
567d5b771a90 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
195
567d5b771a90 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
196 ## 3 ## ONCE we have this better segment (BEST CODING SEGMENT)
567d5b771a90 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
197 ## ==> GET THE STARTING and ENDING POSITIONS (in aa position and in nuc position)
567d5b771a90 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
198 ## And get the INDEX of the best ORF [0, 1, or 2]
567d5b771a90 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
199 if BEST_LONGUEST_SUBSEQUENCE_LIST_POSITION != []:
567d5b771a90 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
200 pos_MIN_aa = BEST_LONGUEST_SUBSEQUENCE_LIST_POSITION[0]
567d5b771a90 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
201 pos_MIN_aa = pos_MIN_aa - 1
567d5b771a90 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
202 pos_MAX_aa = BEST_LONGUEST_SUBSEQUENCE_LIST_POSITION[-1]
567d5b771a90 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
203
567d5b771a90 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
204
567d5b771a90 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
205 BESTORF_bash_of_aligned_aa_seq = {}
567d5b771a90 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
206 BESTORF_bash_of_aligned_aa_seq_CODING = {}
567d5b771a90 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
207 for fasta_name in bash_of_aligned_aa_seq_3ORF.keys():
567d5b771a90 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
208 index_BEST_ORF = BEST_ORF-1 ### cause list going from 0 to 2 in LIST_3_ORF, while the ORF nb is indexed from 1 to 3
567d5b771a90 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
209 seq = bash_of_aligned_aa_seq_3ORF[fasta_name][index_BEST_ORF]
567d5b771a90 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
210 seq_coding = seq[pos_MIN_aa:pos_MAX_aa]
567d5b771a90 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
211 BESTORF_bash_of_aligned_aa_seq[fasta_name] = seq
567d5b771a90 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
212 BESTORF_bash_of_aligned_aa_seq_CODING[fasta_name] = seq_coding
567d5b771a90 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
213
567d5b771a90 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
214 ## 4 ## Get the corresponding position (START/END of BEST CODING SEGMENT) for nucleotides alignment
567d5b771a90 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
215 pos_MIN_nuc = pos_MIN_aa * 3
567d5b771a90 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
216 pos_MAX_nuc = pos_MAX_aa * 3
567d5b771a90 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
217
567d5b771a90 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
218 BESTORF_bash_aligned_nc_seq = {}
2
0d2f72caea10 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit 44a89d5eeb82789bfc643b33c11f391281b6374b
abims-sbr
parents: 1
diff changeset
219 BESTORF_bash_aligned_nc_seq_CODING = {}
1
567d5b771a90 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
220 for fasta_name in bash_aligned_nc_seq.keys():
567d5b771a90 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
221 seq = bash_of_aligned_nuc_seq_3ORF[fasta_name][index_BEST_ORF]
567d5b771a90 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
222 seq_coding = seq[pos_MIN_nuc:pos_MAX_nuc]
567d5b771a90 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
223 BESTORF_bash_aligned_nc_seq[fasta_name] = seq
567d5b771a90 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
224 BESTORF_bash_aligned_nc_seq_CODING[fasta_name] = seq_coding
567d5b771a90 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
225
567d5b771a90 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
226 else: ### no CDS found ###
567d5b771a90 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
227 BESTORF_bash_aligned_nc_seq = {}
567d5b771a90 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
228 BESTORF_bash_aligned_nc_seq_CODING = {}
567d5b771a90 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
229 BESTORF_bash_of_aligned_aa_seq = {}
567d5b771a90 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
230 BESTORF_bash_of_aligned_aa_seq_CODING ={}
567d5b771a90 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
231
567d5b771a90 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
232
567d5b771a90 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
233
567d5b771a90 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
234 ### Check whether their is a "M" or not, and if at least 1 "M" is present, that it is not in the last 50 aa
567d5b771a90 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
235 ###########################################################################################################
567d5b771a90 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
236
567d5b771a90 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
237 BESTORF_bash_of_aligned_aa_seq_CDS_with_M = {}
567d5b771a90 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
238 BESTORF_bash_of_aligned_nuc_seq_CDS_with_M = {}
2
0d2f72caea10 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit 44a89d5eeb82789bfc643b33c11f391281b6374b
abims-sbr
parents: 1
diff changeset
239
1
567d5b771a90 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
240 Ortho = 0
567d5b771a90 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
241 for fasta_name in BESTORF_bash_of_aligned_aa_seq_CODING.keys():
567d5b771a90 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
242 seq_aa = BESTORF_bash_of_aligned_aa_seq_CODING[fasta_name]
567d5b771a90 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
243 Ortho = detect_Methionine(seq_aa, Ortho) ### DEF6 ###
567d5b771a90 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
244
567d5b771a90 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
245 ## CASE 1: A "M" is present and correctly localized (not in last 50 aa)
567d5b771a90 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
246 if Ortho == 1:
567d5b771a90 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
247 BESTORF_bash_of_aligned_aa_seq_CDS_with_M = BESTORF_bash_of_aligned_aa_seq_CODING
567d5b771a90 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
248 BESTORF_bash_of_aligned_nuc_seq_CDS_with_M = BESTORF_bash_aligned_nc_seq_CODING
567d5b771a90 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
249
567d5b771a90 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
250 ## CASE 2: in case the CDS is truncated, so the "M" is maybe missing:
2
0d2f72caea10 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit 44a89d5eeb82789bfc643b33c11f391281b6374b
abims-sbr
parents: 1
diff changeset
251 if Ortho == 0 and CDS_maybe_truncated == 1:
1
567d5b771a90 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
252 BESTORF_bash_of_aligned_aa_seq_CDS_with_M = BESTORF_bash_of_aligned_aa_seq_CODING
567d5b771a90 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
253 BESTORF_bash_of_aligned_nuc_seq_CDS_with_M = BESTORF_bash_aligned_nc_seq_CODING
567d5b771a90 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
254
567d5b771a90 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
255 ## CASE 3: CDS not truncated AND no "M" found in good position (i.e. before the last 50 aa):
567d5b771a90 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
256 ## => the 2 bash "CDS_with_M" are left empty ("{}")
2
0d2f72caea10 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit 44a89d5eeb82789bfc643b33c11f391281b6374b
abims-sbr
parents: 1
diff changeset
257
0d2f72caea10 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit 44a89d5eeb82789bfc643b33c11f391281b6374b
abims-sbr
parents: 1
diff changeset
258 return(BESTORF_bash_aligned_nc_seq, BESTORF_bash_aligned_nc_seq_CODING, BESTORF_bash_of_aligned_nuc_seq_CDS_with_M, BESTORF_bash_of_aligned_aa_seq, BESTORF_bash_of_aligned_aa_seq_CODING, BESTORF_bash_of_aligned_aa_seq_CDS_with_M)
1
567d5b771a90 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
259 ##########################################################
567d5b771a90 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
260
567d5b771a90 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
261
567d5b771a90 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
262 ##################################################################################################
567d5b771a90 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
263 ###### DEF 5 : Detect all indices corresponding to all occurance of a substring in a string ######
567d5b771a90 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
264 ##################################################################################################
567d5b771a90 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
265 def allindices(string, sub):
567d5b771a90 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
266 listindex=[]
567d5b771a90 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
267 offset=0
567d5b771a90 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
268 i = string.find(sub, offset)
567d5b771a90 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
269 while i >= 0:
567d5b771a90 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
270 listindex.append(i)
567d5b771a90 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
271 i = string.find(sub, i + 1)
567d5b771a90 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
272 return listindex
567d5b771a90 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
273 ######################################################
567d5b771a90 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
274
567d5b771a90 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
275
567d5b771a90 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
276 ############################################################
2
0d2f72caea10 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit 44a89d5eeb82789bfc643b33c11f391281b6374b
abims-sbr
parents: 1
diff changeset
277 ###### DEF 6 : Detect if methionin in the aa sequence ######
1
567d5b771a90 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
278 ############################################################
567d5b771a90 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
279 def detect_Methionine(seq_aa, Ortho):
567d5b771a90 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
280
567d5b771a90 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
281 ln = len(seq_aa)
567d5b771a90 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
282 nbre = sys.argv[2]
567d5b771a90 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
283 CUTOFF_Last_50aa = ln - MINIMAL_CDS_LENGTH
567d5b771a90 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
284 #Ortho = 0 ## means orthologs not found
2
0d2f72caea10 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit 44a89d5eeb82789bfc643b33c11f391281b6374b
abims-sbr
parents: 1
diff changeset
285
1
567d5b771a90 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
286 ## Find all indices of occurances of "M" in a string of aa
567d5b771a90 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
287 list_indices = allindices(seq_aa, "M") ### DEF5 ###
2
0d2f72caea10 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit 44a89d5eeb82789bfc643b33c11f391281b6374b
abims-sbr
parents: 1
diff changeset
288
1
567d5b771a90 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
289 ## If some "M" are present, find whether the first "M" found is not in the 50 last aa (indice < CUTOFF_Last_50aa) ==> in this case: maybenot a CDS
567d5b771a90 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
290 if list_indices != []:
2
0d2f72caea10 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit 44a89d5eeb82789bfc643b33c11f391281b6374b
abims-sbr
parents: 1
diff changeset
291 first_M = list_indices[0]
1
567d5b771a90 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
292 if first_M < CUTOFF_Last_50aa:
567d5b771a90 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
293 Ortho = 1 ## means orthologs found
567d5b771a90 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
294
567d5b771a90 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
295 return(Ortho)
567d5b771a90 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
296 ###################################
567d5b771a90 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
297
567d5b771a90 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
298
567d5b771a90 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
299 ############################################################
2
0d2f72caea10 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit 44a89d5eeb82789bfc643b33c11f391281b6374b
abims-sbr
parents: 1
diff changeset
300 ###### DEF 7 : Reverse complement DNA sequence ######
1
567d5b771a90 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
301 ###### Reference: http://crazyhottommy.blogspot.fr/2013/10/python-code-for-getting-reverse.html
567d5b771a90 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
302 ############################################################
567d5b771a90 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
303 def ReverseComplement2(seq):
567d5b771a90 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
304 # too lazy to construct the dictionary manually, use a dict comprehension
9
640ef4c06ed5 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit f1ba8d136e0129f3e8435b25a95f70f697d51464-dirty
abims-sbr
parents: 8
diff changeset
305 seq1 = 'ATCGN-TAGCN-atcgn-tagcn-'
640ef4c06ed5 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit f1ba8d136e0129f3e8435b25a95f70f697d51464-dirty
abims-sbr
parents: 8
diff changeset
306 seq_dict = { seq1[i]:seq1[i+6] for i in range(24) if i < 6 or 12<=i<16 }
1
567d5b771a90 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
307 return "".join([seq_dict[base] for base in reversed(seq)])
8
716a45028e55 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit 90cfcf697b9f128e81bea1270378e59d63ab0a6f
abims-sbr
parents: 7
diff changeset
308 ############################
1
567d5b771a90 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
309
567d5b771a90 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
310
567d5b771a90 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
311 #######################
567d5b771a90 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
312 ##### RUN RUN RUN #####
567d5b771a90 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
313 #######################
567d5b771a90 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
314 import string, os, time, re, zipfile, sys
3
ff98ed7849fa planemo upload for repository https://github.com/abims-sbr/adaptsearch commit cf1b9c905931ca2ca25faa4844d45c908756472f
abims-sbr
parents: 2
diff changeset
315 from dico import dico
1
567d5b771a90 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
316
9
640ef4c06ed5 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit f1ba8d136e0129f3e8435b25a95f70f697d51464-dirty
abims-sbr
parents: 8
diff changeset
317 MINIMAL_CDS_LENGTH = int(sys.argv[2]) ## in aa number
2
0d2f72caea10 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit 44a89d5eeb82789bfc643b33c11f391281b6374b
abims-sbr
parents: 1
diff changeset
318
0d2f72caea10 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit 44a89d5eeb82789bfc643b33c11f391281b6374b
abims-sbr
parents: 1
diff changeset
319 ### Get Universal Code
9
640ef4c06ed5 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit f1ba8d136e0129f3e8435b25a95f70f697d51464-dirty
abims-sbr
parents: 8
diff changeset
320 bash_codeUniversel = code_universel(sys.argv[1]) ### DEF2 ###
640ef4c06ed5 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit f1ba8d136e0129f3e8435b25a95f70f697d51464-dirty
abims-sbr
parents: 8
diff changeset
321
640ef4c06ed5 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit f1ba8d136e0129f3e8435b25a95f70f697d51464-dirty
abims-sbr
parents: 8
diff changeset
322 ## INPUT from file containing list of species
640ef4c06ed5 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit f1ba8d136e0129f3e8435b25a95f70f697d51464-dirty
abims-sbr
parents: 8
diff changeset
323 list_files = []
640ef4c06ed5 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit f1ba8d136e0129f3e8435b25a95f70f697d51464-dirty
abims-sbr
parents: 8
diff changeset
324 with open(sys.argv[3], 'r') as f:
640ef4c06ed5 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit f1ba8d136e0129f3e8435b25a95f70f697d51464-dirty
abims-sbr
parents: 8
diff changeset
325 for line in f.readlines():
640ef4c06ed5 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit f1ba8d136e0129f3e8435b25a95f70f697d51464-dirty
abims-sbr
parents: 8
diff changeset
326 list_files.append(line.strip('\n'))
1
567d5b771a90 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
327
567d5b771a90 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
328 os.mkdir("04_BEST_ORF_nuc")
567d5b771a90 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
329 Path_OUT1 = "04_BEST_ORF_nuc"
567d5b771a90 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
330 os.mkdir("04_BEST_ORF_aa")
567d5b771a90 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
331 Path_OUT2 = "04_BEST_ORF_aa"
567d5b771a90 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
332
567d5b771a90 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
333 os.mkdir("05_CDS_nuc")
567d5b771a90 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
334 Path_OUT3 = "05_CDS_nuc"
567d5b771a90 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
335 os.mkdir("05_CDS_aa")
567d5b771a90 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
336 Path_OUT4 = "05_CDS_aa"
567d5b771a90 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
337
567d5b771a90 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
338 os.mkdir("06_CDS_with_M_nuc")
567d5b771a90 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
339 Path_OUT5 = "06_CDS_with_M_nuc"
567d5b771a90 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
340 os.mkdir("06_CDS_with_M_aa")
567d5b771a90 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
341 Path_OUT6 = "06_CDS_with_M_aa"
567d5b771a90 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
342
567d5b771a90 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
343 ### Get the Bash corresponding to an alignment file in fasta format
567d5b771a90 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
344 count_file_processed = 0
567d5b771a90 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
345 count_file_with_CDS = 0
567d5b771a90 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
346 count_file_without_CDS = 0
567d5b771a90 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
347 count_file_with_CDS_plus_M = 0
567d5b771a90 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
348
8
716a45028e55 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit 90cfcf697b9f128e81bea1270378e59d63ab0a6f
abims-sbr
parents: 7
diff changeset
349 # ! : Currently, files are named "Orthogroup_x_y_sequences.fasta, where x is the number of the orthogroup (not important, juste here to make a distinct name),
716a45028e55 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit 90cfcf697b9f128e81bea1270378e59d63ab0a6f
abims-sbr
parents: 7
diff changeset
350 # and y is the number of sequences/species in the group. These files are outputs of blastalign, where species can be removed. y is then modified.
716a45028e55 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit 90cfcf697b9f128e81bea1270378e59d63ab0a6f
abims-sbr
parents: 7
diff changeset
351
716a45028e55 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit 90cfcf697b9f128e81bea1270378e59d63ab0a6f
abims-sbr
parents: 7
diff changeset
352 name_elems = ["orthogroup", "0", "with", "0", "species.fasta"]
716a45028e55 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit 90cfcf697b9f128e81bea1270378e59d63ab0a6f
abims-sbr
parents: 7
diff changeset
353
716a45028e55 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit 90cfcf697b9f128e81bea1270378e59d63ab0a6f
abims-sbr
parents: 7
diff changeset
354 # by fixing the counter here, there will be some "holes" in the outputs directories (missing numbers), but the groups between directories will correspond
9
640ef4c06ed5 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit f1ba8d136e0129f3e8435b25a95f70f697d51464-dirty
abims-sbr
parents: 8
diff changeset
355 #n0 = 0
640ef4c06ed5 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit f1ba8d136e0129f3e8435b25a95f70f697d51464-dirty
abims-sbr
parents: 8
diff changeset
356
640ef4c06ed5 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit f1ba8d136e0129f3e8435b25a95f70f697d51464-dirty
abims-sbr
parents: 8
diff changeset
357 for file in list_files:
640ef4c06ed5 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit f1ba8d136e0129f3e8435b25a95f70f697d51464-dirty
abims-sbr
parents: 8
diff changeset
358 #n0 += 1
8
716a45028e55 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit 90cfcf697b9f128e81bea1270378e59d63ab0a6f
abims-sbr
parents: 7
diff changeset
359
2
0d2f72caea10 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit 44a89d5eeb82789bfc643b33c11f391281b6374b
abims-sbr
parents: 1
diff changeset
360 count_file_processed = count_file_processed + 1
9
640ef4c06ed5 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit f1ba8d136e0129f3e8435b25a95f70f697d51464-dirty
abims-sbr
parents: 8
diff changeset
361 nb_gp = file.split('_')[1] # Keep trace of the orthogroup number
7
35e39b4128ba planemo upload for repository https://github.com/abims-sbr/adaptsearch commit b7a3030ea134b5dfad89b1a869db659d72d1145c
abims-sbr
parents: 3
diff changeset
362 fasta_file_path = "./%s" %file
35e39b4128ba planemo upload for repository https://github.com/abims-sbr/adaptsearch commit b7a3030ea134b5dfad89b1a869db659d72d1145c
abims-sbr
parents: 3
diff changeset
363 bash_fasta = dico(fasta_file_path) ### DEF 1 ###
2
0d2f72caea10 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit 44a89d5eeb82789bfc643b33c11f391281b6374b
abims-sbr
parents: 1
diff changeset
364 BESTORF_nuc, BESTORF_nuc_CODING, BESTORF_nuc_CDS_with_M, BESTORF_aa, BESTORF_aa_CODING, BESTORF_aa_CDS_with_M = find_good_ORF_criteria_3(bash_fasta, bash_codeUniversel) ### DEF 4 - PART 2 - ###
8
716a45028e55 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit 90cfcf697b9f128e81bea1270378e59d63ab0a6f
abims-sbr
parents: 7
diff changeset
365
9
640ef4c06ed5 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit f1ba8d136e0129f3e8435b25a95f70f697d51464-dirty
abims-sbr
parents: 8
diff changeset
366 name_elems[1] = nb_gp
1
567d5b771a90 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
367
567d5b771a90 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
368 ## a ## OUTPUT BESTORF_nuc
567d5b771a90 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
369 if BESTORF_nuc != {}:
8
716a45028e55 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit 90cfcf697b9f128e81bea1270378e59d63ab0a6f
abims-sbr
parents: 7
diff changeset
370 name_elems[3] = str(len(BESTORF_nuc.keys()))
716a45028e55 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit 90cfcf697b9f128e81bea1270378e59d63ab0a6f
abims-sbr
parents: 7
diff changeset
371 new_name = "_".join(name_elems)
1
567d5b771a90 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
372 count_file_with_CDS = count_file_with_CDS +1
8
716a45028e55 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit 90cfcf697b9f128e81bea1270378e59d63ab0a6f
abims-sbr
parents: 7
diff changeset
373 OUT1 = open("%s/%s" %(Path_OUT1,new_name), "w")
1
567d5b771a90 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
374 for fasta_name in BESTORF_nuc.keys():
567d5b771a90 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
375 seq = BESTORF_nuc[fasta_name]
567d5b771a90 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
376 OUT1.write("%s\n" %fasta_name)
567d5b771a90 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
377 OUT1.write("%s\n" %seq)
567d5b771a90 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
378 OUT1.close()
567d5b771a90 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
379 else:
567d5b771a90 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
380 count_file_without_CDS = count_file_without_CDS + 1
2
0d2f72caea10 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit 44a89d5eeb82789bfc643b33c11f391281b6374b
abims-sbr
parents: 1
diff changeset
381
1
567d5b771a90 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
382 ## b ## OUTPUT BESTORF_nuc_CODING ===> THE MOST INTERESTING!!!
567d5b771a90 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
383 if BESTORF_aa != {}:
8
716a45028e55 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit 90cfcf697b9f128e81bea1270378e59d63ab0a6f
abims-sbr
parents: 7
diff changeset
384 name_elems[3] = str(len(BESTORF_aa.keys()))
716a45028e55 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit 90cfcf697b9f128e81bea1270378e59d63ab0a6f
abims-sbr
parents: 7
diff changeset
385 new_name = "_".join(name_elems)
716a45028e55 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit 90cfcf697b9f128e81bea1270378e59d63ab0a6f
abims-sbr
parents: 7
diff changeset
386 OUT2 = open("%s/%s" %(Path_OUT2,new_name), "w")
1
567d5b771a90 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
387 for fasta_name in BESTORF_aa.keys():
567d5b771a90 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
388 seq = BESTORF_aa[fasta_name]
567d5b771a90 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
389 OUT2.write("%s\n" %fasta_name)
567d5b771a90 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
390 OUT2.write("%s\n" %seq)
2
0d2f72caea10 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit 44a89d5eeb82789bfc643b33c11f391281b6374b
abims-sbr
parents: 1
diff changeset
391 OUT2.close()
1
567d5b771a90 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
392
567d5b771a90 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
393 ## c ## OUTPUT BESTORF_aa
567d5b771a90 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
394 if BESTORF_nuc_CODING != {}:
8
716a45028e55 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit 90cfcf697b9f128e81bea1270378e59d63ab0a6f
abims-sbr
parents: 7
diff changeset
395 name_elems[3] = str(len(BESTORF_nuc_CODING.keys()))
716a45028e55 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit 90cfcf697b9f128e81bea1270378e59d63ab0a6f
abims-sbr
parents: 7
diff changeset
396 new_name = "_".join(name_elems)
716a45028e55 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit 90cfcf697b9f128e81bea1270378e59d63ab0a6f
abims-sbr
parents: 7
diff changeset
397 OUT3 = open("%s/%s" %(Path_OUT3,new_name), "w")
1
567d5b771a90 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
398 for fasta_name in BESTORF_nuc_CODING.keys():
567d5b771a90 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
399 seq = BESTORF_nuc_CODING[fasta_name]
567d5b771a90 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
400 OUT3.write("%s\n" %fasta_name)
567d5b771a90 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
401 OUT3.write("%s\n" %seq)
567d5b771a90 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
402 OUT3.close()
567d5b771a90 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
403
567d5b771a90 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
404 ## d ## OUTPUT BESTORF_aa_CODING
567d5b771a90 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
405 if BESTORF_aa_CODING != {}:
8
716a45028e55 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit 90cfcf697b9f128e81bea1270378e59d63ab0a6f
abims-sbr
parents: 7
diff changeset
406 name_elems[3] = str(len(BESTORF_aa_CODING.keys()))
716a45028e55 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit 90cfcf697b9f128e81bea1270378e59d63ab0a6f
abims-sbr
parents: 7
diff changeset
407 new_name = "_".join(name_elems)
716a45028e55 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit 90cfcf697b9f128e81bea1270378e59d63ab0a6f
abims-sbr
parents: 7
diff changeset
408 OUT4 = open("%s/%s" %(Path_OUT4,new_name), "w")
1
567d5b771a90 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
409 for fasta_name in BESTORF_aa_CODING.keys():
567d5b771a90 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
410 seq = BESTORF_aa_CODING[fasta_name]
567d5b771a90 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
411 OUT4.write("%s\n" %fasta_name)
567d5b771a90 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
412 OUT4.write("%s\n" %seq)
2
0d2f72caea10 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit 44a89d5eeb82789bfc643b33c11f391281b6374b
abims-sbr
parents: 1
diff changeset
413 OUT4.close()
1
567d5b771a90 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
414
567d5b771a90 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
415 ## e ## OUTPUT BESTORF_nuc_CDS_with_M
567d5b771a90 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
416 if BESTORF_nuc_CDS_with_M != {}:
8
716a45028e55 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit 90cfcf697b9f128e81bea1270378e59d63ab0a6f
abims-sbr
parents: 7
diff changeset
417 name_elems[3] = str(len(BESTORF_nuc_CDS_with_M.keys()))
716a45028e55 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit 90cfcf697b9f128e81bea1270378e59d63ab0a6f
abims-sbr
parents: 7
diff changeset
418 new_name = "_".join(name_elems)
1
567d5b771a90 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
419 count_file_with_CDS_plus_M = count_file_with_CDS_plus_M + 1
8
716a45028e55 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit 90cfcf697b9f128e81bea1270378e59d63ab0a6f
abims-sbr
parents: 7
diff changeset
420 OUT5 = open("%s/%s" %(Path_OUT5,new_name), "w")
1
567d5b771a90 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
421 for fasta_name in BESTORF_nuc_CDS_with_M.keys():
567d5b771a90 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
422 seq = BESTORF_nuc_CDS_with_M[fasta_name]
567d5b771a90 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
423 OUT5.write("%s\n" %fasta_name)
567d5b771a90 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
424 OUT5.write("%s\n" %seq)
2
0d2f72caea10 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit 44a89d5eeb82789bfc643b33c11f391281b6374b
abims-sbr
parents: 1
diff changeset
425 OUT5.close()
1
567d5b771a90 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
426
567d5b771a90 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
427 ## f ## OUTPUT BESTORF_aa_CDS_with_M
567d5b771a90 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
428 if BESTORF_aa_CDS_with_M != {}:
8
716a45028e55 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit 90cfcf697b9f128e81bea1270378e59d63ab0a6f
abims-sbr
parents: 7
diff changeset
429 name_elems[3] = str(len(BESTORF_aa_CDS_with_M.keys()))
716a45028e55 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit 90cfcf697b9f128e81bea1270378e59d63ab0a6f
abims-sbr
parents: 7
diff changeset
430 new_name = "_".join(name_elems)
716a45028e55 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit 90cfcf697b9f128e81bea1270378e59d63ab0a6f
abims-sbr
parents: 7
diff changeset
431 OUT6 = open("%s/%s" %(Path_OUT6,new_name), "w")
1
567d5b771a90 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
432 for fasta_name in BESTORF_aa_CDS_with_M.keys():
567d5b771a90 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
433 seq = BESTORF_aa_CDS_with_M[fasta_name]
567d5b771a90 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
434 OUT6.write("%s\n" %fasta_name)
567d5b771a90 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
435 OUT6.write("%s\n" %seq)
567d5b771a90 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
436 OUT6.close()
567d5b771a90 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
437
9
640ef4c06ed5 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit f1ba8d136e0129f3e8435b25a95f70f697d51464-dirty
abims-sbr
parents: 8
diff changeset
438 #os.system("rm -rf %s" %file)
1
567d5b771a90 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
439
2
0d2f72caea10 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit 44a89d5eeb82789bfc643b33c11f391281b6374b
abims-sbr
parents: 1
diff changeset
440 ## Print
1
567d5b771a90 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
441 print "*************** CDS detection ***************"
567d5b771a90 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
442 print "\nFiles processed: %d" %count_file_processed
567d5b771a90 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
443 print "\tFiles with CDS: %d" %count_file_with_CDS
567d5b771a90 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
444 print "\t\tFiles with CDS plus M (codon start): %d" %count_file_with_CDS_plus_M
567d5b771a90 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
445 print "\tFiles without CDS: %d \n" %count_file_without_CDS
567d5b771a90 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
446 print ""