annotate scripts/S03_remove_site_with_not_enough_species_represented.py @ 0:eb95bf7f90ae draft

planemo upload for repository https://github.com/abims-sbr/adaptsearch commit 3c7982d775b6f3b472f6514d791edcb43cd258a1-dirty
author abims-sbr
date Fri, 01 Feb 2019 10:26:37 -0500
parents
children c79bdda8abfb
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
rev   line source
0
eb95bf7f90ae planemo upload for repository https://github.com/abims-sbr/adaptsearch commit 3c7982d775b6f3b472f6514d791edcb43cd258a1-dirty
abims-sbr
parents:
diff changeset
1 #!/usr/bin/env python
eb95bf7f90ae planemo upload for repository https://github.com/abims-sbr/adaptsearch commit 3c7982d775b6f3b472f6514d791edcb43cd258a1-dirty
abims-sbr
parents:
diff changeset
2 # coding: utf8
eb95bf7f90ae planemo upload for repository https://github.com/abims-sbr/adaptsearch commit 3c7982d775b6f3b472f6514d791edcb43cd258a1-dirty
abims-sbr
parents:
diff changeset
3 ## Author: Eric Fontanillas
eb95bf7f90ae planemo upload for repository https://github.com/abims-sbr/adaptsearch commit 3c7982d775b6f3b472f6514d791edcb43cd258a1-dirty
abims-sbr
parents:
diff changeset
4 ## Modification: 03/09/14 by Julie BAFFARD
eb95bf7f90ae planemo upload for repository https://github.com/abims-sbr/adaptsearch commit 3c7982d775b6f3b472f6514d791edcb43cd258a1-dirty
abims-sbr
parents:
diff changeset
5 ## Last modification : 05/03/18 by Victor Mataigne
eb95bf7f90ae planemo upload for repository https://github.com/abims-sbr/adaptsearch commit 3c7982d775b6f3b472f6514d791edcb43cd258a1-dirty
abims-sbr
parents:
diff changeset
6
eb95bf7f90ae planemo upload for repository https://github.com/abims-sbr/adaptsearch commit 3c7982d775b6f3b472f6514d791edcb43cd258a1-dirty
abims-sbr
parents:
diff changeset
7 ## Description : find and remove indels
eb95bf7f90ae planemo upload for repository https://github.com/abims-sbr/adaptsearch commit 3c7982d775b6f3b472f6514d791edcb43cd258a1-dirty
abims-sbr
parents:
diff changeset
8
eb95bf7f90ae planemo upload for repository https://github.com/abims-sbr/adaptsearch commit 3c7982d775b6f3b472f6514d791edcb43cd258a1-dirty
abims-sbr
parents:
diff changeset
9 ####################
eb95bf7f90ae planemo upload for repository https://github.com/abims-sbr/adaptsearch commit 3c7982d775b6f3b472f6514d791edcb43cd258a1-dirty
abims-sbr
parents:
diff changeset
10 ###### DEF 2 #######
eb95bf7f90ae planemo upload for repository https://github.com/abims-sbr/adaptsearch commit 3c7982d775b6f3b472f6514d791edcb43cd258a1-dirty
abims-sbr
parents:
diff changeset
11 ####################
eb95bf7f90ae planemo upload for repository https://github.com/abims-sbr/adaptsearch commit 3c7982d775b6f3b472f6514d791edcb43cd258a1-dirty
abims-sbr
parents:
diff changeset
12 def remove_position_with_too_much_missing_data(bash_aa, bash_nuc, MIN_SPECIES_NB):
eb95bf7f90ae planemo upload for repository https://github.com/abims-sbr/adaptsearch commit 3c7982d775b6f3b472f6514d791edcb43cd258a1-dirty
abims-sbr
parents:
diff changeset
13
eb95bf7f90ae planemo upload for repository https://github.com/abims-sbr/adaptsearch commit 3c7982d775b6f3b472f6514d791edcb43cd258a1-dirty
abims-sbr
parents:
diff changeset
14 ## 1 ## Get alignment length
eb95bf7f90ae planemo upload for repository https://github.com/abims-sbr/adaptsearch commit 3c7982d775b6f3b472f6514d791edcb43cd258a1-dirty
abims-sbr
parents:
diff changeset
15 fasta_name0 = bash_aa.keys()[0]
eb95bf7f90ae planemo upload for repository https://github.com/abims-sbr/adaptsearch commit 3c7982d775b6f3b472f6514d791edcb43cd258a1-dirty
abims-sbr
parents:
diff changeset
16 ln_aa = len(bash_aa[fasta_name0])
eb95bf7f90ae planemo upload for repository https://github.com/abims-sbr/adaptsearch commit 3c7982d775b6f3b472f6514d791edcb43cd258a1-dirty
abims-sbr
parents:
diff changeset
17
eb95bf7f90ae planemo upload for repository https://github.com/abims-sbr/adaptsearch commit 3c7982d775b6f3b472f6514d791edcb43cd258a1-dirty
abims-sbr
parents:
diff changeset
18 ln_nuc = len(bash_nuc[fasta_name0])
eb95bf7f90ae planemo upload for repository https://github.com/abims-sbr/adaptsearch commit 3c7982d775b6f3b472f6514d791edcb43cd258a1-dirty
abims-sbr
parents:
diff changeset
19
eb95bf7f90ae planemo upload for repository https://github.com/abims-sbr/adaptsearch commit 3c7982d775b6f3b472f6514d791edcb43cd258a1-dirty
abims-sbr
parents:
diff changeset
20
eb95bf7f90ae planemo upload for repository https://github.com/abims-sbr/adaptsearch commit 3c7982d775b6f3b472f6514d791edcb43cd258a1-dirty
abims-sbr
parents:
diff changeset
21 ## 2 ## Get positions keeped in aa alignment
eb95bf7f90ae planemo upload for repository https://github.com/abims-sbr/adaptsearch commit 3c7982d775b6f3b472f6514d791edcb43cd258a1-dirty
abims-sbr
parents:
diff changeset
22 LIST_POSITION_KEEPED_aa = []
eb95bf7f90ae planemo upload for repository https://github.com/abims-sbr/adaptsearch commit 3c7982d775b6f3b472f6514d791edcb43cd258a1-dirty
abims-sbr
parents:
diff changeset
23 i=0
eb95bf7f90ae planemo upload for repository https://github.com/abims-sbr/adaptsearch commit 3c7982d775b6f3b472f6514d791edcb43cd258a1-dirty
abims-sbr
parents:
diff changeset
24 while i < ln_aa:
eb95bf7f90ae planemo upload for repository https://github.com/abims-sbr/adaptsearch commit 3c7982d775b6f3b472f6514d791edcb43cd258a1-dirty
abims-sbr
parents:
diff changeset
25 site = []
eb95bf7f90ae planemo upload for repository https://github.com/abims-sbr/adaptsearch commit 3c7982d775b6f3b472f6514d791edcb43cd258a1-dirty
abims-sbr
parents:
diff changeset
26 for fasta_name in bash_aa.keys():
eb95bf7f90ae planemo upload for repository https://github.com/abims-sbr/adaptsearch commit 3c7982d775b6f3b472f6514d791edcb43cd258a1-dirty
abims-sbr
parents:
diff changeset
27 pos = bash_aa[fasta_name][i]
eb95bf7f90ae planemo upload for repository https://github.com/abims-sbr/adaptsearch commit 3c7982d775b6f3b472f6514d791edcb43cd258a1-dirty
abims-sbr
parents:
diff changeset
28
eb95bf7f90ae planemo upload for repository https://github.com/abims-sbr/adaptsearch commit 3c7982d775b6f3b472f6514d791edcb43cd258a1-dirty
abims-sbr
parents:
diff changeset
29 if pos != "-" and pos != "?" and pos != "X":
eb95bf7f90ae planemo upload for repository https://github.com/abims-sbr/adaptsearch commit 3c7982d775b6f3b472f6514d791edcb43cd258a1-dirty
abims-sbr
parents:
diff changeset
30 site.append(pos)
eb95bf7f90ae planemo upload for repository https://github.com/abims-sbr/adaptsearch commit 3c7982d775b6f3b472f6514d791edcb43cd258a1-dirty
abims-sbr
parents:
diff changeset
31 if len(site) >= MIN_SPECIES_NB:
eb95bf7f90ae planemo upload for repository https://github.com/abims-sbr/adaptsearch commit 3c7982d775b6f3b472f6514d791edcb43cd258a1-dirty
abims-sbr
parents:
diff changeset
32 LIST_POSITION_KEEPED_aa.append(i)
eb95bf7f90ae planemo upload for repository https://github.com/abims-sbr/adaptsearch commit 3c7982d775b6f3b472f6514d791edcb43cd258a1-dirty
abims-sbr
parents:
diff changeset
33 i = i+1
eb95bf7f90ae planemo upload for repository https://github.com/abims-sbr/adaptsearch commit 3c7982d775b6f3b472f6514d791edcb43cd258a1-dirty
abims-sbr
parents:
diff changeset
34
eb95bf7f90ae planemo upload for repository https://github.com/abims-sbr/adaptsearch commit 3c7982d775b6f3b472f6514d791edcb43cd258a1-dirty
abims-sbr
parents:
diff changeset
35 ## 3 ## Get positions keeped in nuc alignment
eb95bf7f90ae planemo upload for repository https://github.com/abims-sbr/adaptsearch commit 3c7982d775b6f3b472f6514d791edcb43cd258a1-dirty
abims-sbr
parents:
diff changeset
36 LIST_POSITION_KEEPED_nuc = []
eb95bf7f90ae planemo upload for repository https://github.com/abims-sbr/adaptsearch commit 3c7982d775b6f3b472f6514d791edcb43cd258a1-dirty
abims-sbr
parents:
diff changeset
37 for position in LIST_POSITION_KEEPED_aa:
eb95bf7f90ae planemo upload for repository https://github.com/abims-sbr/adaptsearch commit 3c7982d775b6f3b472f6514d791edcb43cd258a1-dirty
abims-sbr
parents:
diff changeset
38 position1 = position*3
eb95bf7f90ae planemo upload for repository https://github.com/abims-sbr/adaptsearch commit 3c7982d775b6f3b472f6514d791edcb43cd258a1-dirty
abims-sbr
parents:
diff changeset
39 position2 = position*3 + 1
eb95bf7f90ae planemo upload for repository https://github.com/abims-sbr/adaptsearch commit 3c7982d775b6f3b472f6514d791edcb43cd258a1-dirty
abims-sbr
parents:
diff changeset
40 position3 = position*3 + 2
eb95bf7f90ae planemo upload for repository https://github.com/abims-sbr/adaptsearch commit 3c7982d775b6f3b472f6514d791edcb43cd258a1-dirty
abims-sbr
parents:
diff changeset
41 LIST_POSITION_KEEPED_nuc.append(position1)
eb95bf7f90ae planemo upload for repository https://github.com/abims-sbr/adaptsearch commit 3c7982d775b6f3b472f6514d791edcb43cd258a1-dirty
abims-sbr
parents:
diff changeset
42 LIST_POSITION_KEEPED_nuc.append(position2)
eb95bf7f90ae planemo upload for repository https://github.com/abims-sbr/adaptsearch commit 3c7982d775b6f3b472f6514d791edcb43cd258a1-dirty
abims-sbr
parents:
diff changeset
43 LIST_POSITION_KEEPED_nuc.append(position3)
eb95bf7f90ae planemo upload for repository https://github.com/abims-sbr/adaptsearch commit 3c7982d775b6f3b472f6514d791edcb43cd258a1-dirty
abims-sbr
parents:
diff changeset
44
eb95bf7f90ae planemo upload for repository https://github.com/abims-sbr/adaptsearch commit 3c7982d775b6f3b472f6514d791edcb43cd258a1-dirty
abims-sbr
parents:
diff changeset
45 ## 4 ## Create entries for "filtered_bash" for aa & nuc
eb95bf7f90ae planemo upload for repository https://github.com/abims-sbr/adaptsearch commit 3c7982d775b6f3b472f6514d791edcb43cd258a1-dirty
abims-sbr
parents:
diff changeset
46 filtered_bash_aa = {}
eb95bf7f90ae planemo upload for repository https://github.com/abims-sbr/adaptsearch commit 3c7982d775b6f3b472f6514d791edcb43cd258a1-dirty
abims-sbr
parents:
diff changeset
47 filtered_bash_nuc = {}
eb95bf7f90ae planemo upload for repository https://github.com/abims-sbr/adaptsearch commit 3c7982d775b6f3b472f6514d791edcb43cd258a1-dirty
abims-sbr
parents:
diff changeset
48 for fasta_name in bash_aa.keys():
eb95bf7f90ae planemo upload for repository https://github.com/abims-sbr/adaptsearch commit 3c7982d775b6f3b472f6514d791edcb43cd258a1-dirty
abims-sbr
parents:
diff changeset
49 filtered_bash_aa[fasta_name] = ""
eb95bf7f90ae planemo upload for repository https://github.com/abims-sbr/adaptsearch commit 3c7982d775b6f3b472f6514d791edcb43cd258a1-dirty
abims-sbr
parents:
diff changeset
50 for fasta_name in bash_nuc.keys():
eb95bf7f90ae planemo upload for repository https://github.com/abims-sbr/adaptsearch commit 3c7982d775b6f3b472f6514d791edcb43cd258a1-dirty
abims-sbr
parents:
diff changeset
51 filtered_bash_nuc[fasta_name] = ""
eb95bf7f90ae planemo upload for repository https://github.com/abims-sbr/adaptsearch commit 3c7982d775b6f3b472f6514d791edcb43cd258a1-dirty
abims-sbr
parents:
diff changeset
52
eb95bf7f90ae planemo upload for repository https://github.com/abims-sbr/adaptsearch commit 3c7982d775b6f3b472f6514d791edcb43cd258a1-dirty
abims-sbr
parents:
diff changeset
53 ## 5 ## Write "filtered_bash" for aa
eb95bf7f90ae planemo upload for repository https://github.com/abims-sbr/adaptsearch commit 3c7982d775b6f3b472f6514d791edcb43cd258a1-dirty
abims-sbr
parents:
diff changeset
54 j=0
eb95bf7f90ae planemo upload for repository https://github.com/abims-sbr/adaptsearch commit 3c7982d775b6f3b472f6514d791edcb43cd258a1-dirty
abims-sbr
parents:
diff changeset
55 while j < ln_aa:
eb95bf7f90ae planemo upload for repository https://github.com/abims-sbr/adaptsearch commit 3c7982d775b6f3b472f6514d791edcb43cd258a1-dirty
abims-sbr
parents:
diff changeset
56 for fasta_name in bash_aa.keys():
eb95bf7f90ae planemo upload for repository https://github.com/abims-sbr/adaptsearch commit 3c7982d775b6f3b472f6514d791edcb43cd258a1-dirty
abims-sbr
parents:
diff changeset
57 seq=filtered_bash_aa[fasta_name]
eb95bf7f90ae planemo upload for repository https://github.com/abims-sbr/adaptsearch commit 3c7982d775b6f3b472f6514d791edcb43cd258a1-dirty
abims-sbr
parents:
diff changeset
58 pos=bash_aa[fasta_name][j]
eb95bf7f90ae planemo upload for repository https://github.com/abims-sbr/adaptsearch commit 3c7982d775b6f3b472f6514d791edcb43cd258a1-dirty
abims-sbr
parents:
diff changeset
59
eb95bf7f90ae planemo upload for repository https://github.com/abims-sbr/adaptsearch commit 3c7982d775b6f3b472f6514d791edcb43cd258a1-dirty
abims-sbr
parents:
diff changeset
60 if j in LIST_POSITION_KEEPED_aa:
eb95bf7f90ae planemo upload for repository https://github.com/abims-sbr/adaptsearch commit 3c7982d775b6f3b472f6514d791edcb43cd258a1-dirty
abims-sbr
parents:
diff changeset
61 seq = seq + pos
eb95bf7f90ae planemo upload for repository https://github.com/abims-sbr/adaptsearch commit 3c7982d775b6f3b472f6514d791edcb43cd258a1-dirty
abims-sbr
parents:
diff changeset
62 filtered_bash_aa[fasta_name] = seq
eb95bf7f90ae planemo upload for repository https://github.com/abims-sbr/adaptsearch commit 3c7982d775b6f3b472f6514d791edcb43cd258a1-dirty
abims-sbr
parents:
diff changeset
63 j = j + 1
eb95bf7f90ae planemo upload for repository https://github.com/abims-sbr/adaptsearch commit 3c7982d775b6f3b472f6514d791edcb43cd258a1-dirty
abims-sbr
parents:
diff changeset
64
eb95bf7f90ae planemo upload for repository https://github.com/abims-sbr/adaptsearch commit 3c7982d775b6f3b472f6514d791edcb43cd258a1-dirty
abims-sbr
parents:
diff changeset
65 ## 6 ## Remove empty sequence
eb95bf7f90ae planemo upload for repository https://github.com/abims-sbr/adaptsearch commit 3c7982d775b6f3b472f6514d791edcb43cd258a1-dirty
abims-sbr
parents:
diff changeset
66 for name in filtered_bash_aa.keys():
eb95bf7f90ae planemo upload for repository https://github.com/abims-sbr/adaptsearch commit 3c7982d775b6f3b472f6514d791edcb43cd258a1-dirty
abims-sbr
parents:
diff changeset
67 seq = filtered_bash_aa[name]
eb95bf7f90ae planemo upload for repository https://github.com/abims-sbr/adaptsearch commit 3c7982d775b6f3b472f6514d791edcb43cd258a1-dirty
abims-sbr
parents:
diff changeset
68 if seq == '':
eb95bf7f90ae planemo upload for repository https://github.com/abims-sbr/adaptsearch commit 3c7982d775b6f3b472f6514d791edcb43cd258a1-dirty
abims-sbr
parents:
diff changeset
69 del filtered_bash_aa[name]
eb95bf7f90ae planemo upload for repository https://github.com/abims-sbr/adaptsearch commit 3c7982d775b6f3b472f6514d791edcb43cd258a1-dirty
abims-sbr
parents:
diff changeset
70
eb95bf7f90ae planemo upload for repository https://github.com/abims-sbr/adaptsearch commit 3c7982d775b6f3b472f6514d791edcb43cd258a1-dirty
abims-sbr
parents:
diff changeset
71
eb95bf7f90ae planemo upload for repository https://github.com/abims-sbr/adaptsearch commit 3c7982d775b6f3b472f6514d791edcb43cd258a1-dirty
abims-sbr
parents:
diff changeset
72 ## 7 ## Write "filtered_bash" for nuc
eb95bf7f90ae planemo upload for repository https://github.com/abims-sbr/adaptsearch commit 3c7982d775b6f3b472f6514d791edcb43cd258a1-dirty
abims-sbr
parents:
diff changeset
73 j=0
eb95bf7f90ae planemo upload for repository https://github.com/abims-sbr/adaptsearch commit 3c7982d775b6f3b472f6514d791edcb43cd258a1-dirty
abims-sbr
parents:
diff changeset
74 while j < ln_nuc:
eb95bf7f90ae planemo upload for repository https://github.com/abims-sbr/adaptsearch commit 3c7982d775b6f3b472f6514d791edcb43cd258a1-dirty
abims-sbr
parents:
diff changeset
75 for fasta_name in bash_nuc.keys():
eb95bf7f90ae planemo upload for repository https://github.com/abims-sbr/adaptsearch commit 3c7982d775b6f3b472f6514d791edcb43cd258a1-dirty
abims-sbr
parents:
diff changeset
76 seq=filtered_bash_nuc[fasta_name]
eb95bf7f90ae planemo upload for repository https://github.com/abims-sbr/adaptsearch commit 3c7982d775b6f3b472f6514d791edcb43cd258a1-dirty
abims-sbr
parents:
diff changeset
77 #print seq
eb95bf7f90ae planemo upload for repository https://github.com/abims-sbr/adaptsearch commit 3c7982d775b6f3b472f6514d791edcb43cd258a1-dirty
abims-sbr
parents:
diff changeset
78 pos=bash_nuc[fasta_name][j]
eb95bf7f90ae planemo upload for repository https://github.com/abims-sbr/adaptsearch commit 3c7982d775b6f3b472f6514d791edcb43cd258a1-dirty
abims-sbr
parents:
diff changeset
79
eb95bf7f90ae planemo upload for repository https://github.com/abims-sbr/adaptsearch commit 3c7982d775b6f3b472f6514d791edcb43cd258a1-dirty
abims-sbr
parents:
diff changeset
80 if j in LIST_POSITION_KEEPED_nuc:
eb95bf7f90ae planemo upload for repository https://github.com/abims-sbr/adaptsearch commit 3c7982d775b6f3b472f6514d791edcb43cd258a1-dirty
abims-sbr
parents:
diff changeset
81 seq = seq + pos
eb95bf7f90ae planemo upload for repository https://github.com/abims-sbr/adaptsearch commit 3c7982d775b6f3b472f6514d791edcb43cd258a1-dirty
abims-sbr
parents:
diff changeset
82 filtered_bash_nuc[fasta_name] = seq
eb95bf7f90ae planemo upload for repository https://github.com/abims-sbr/adaptsearch commit 3c7982d775b6f3b472f6514d791edcb43cd258a1-dirty
abims-sbr
parents:
diff changeset
83 j = j + 1
eb95bf7f90ae planemo upload for repository https://github.com/abims-sbr/adaptsearch commit 3c7982d775b6f3b472f6514d791edcb43cd258a1-dirty
abims-sbr
parents:
diff changeset
84
eb95bf7f90ae planemo upload for repository https://github.com/abims-sbr/adaptsearch commit 3c7982d775b6f3b472f6514d791edcb43cd258a1-dirty
abims-sbr
parents:
diff changeset
85 ## 8 ## Remove empty sequence
eb95bf7f90ae planemo upload for repository https://github.com/abims-sbr/adaptsearch commit 3c7982d775b6f3b472f6514d791edcb43cd258a1-dirty
abims-sbr
parents:
diff changeset
86 for name in filtered_bash_nuc.keys():
eb95bf7f90ae planemo upload for repository https://github.com/abims-sbr/adaptsearch commit 3c7982d775b6f3b472f6514d791edcb43cd258a1-dirty
abims-sbr
parents:
diff changeset
87 seq = filtered_bash_nuc[name]
eb95bf7f90ae planemo upload for repository https://github.com/abims-sbr/adaptsearch commit 3c7982d775b6f3b472f6514d791edcb43cd258a1-dirty
abims-sbr
parents:
diff changeset
88 if seq == '':
eb95bf7f90ae planemo upload for repository https://github.com/abims-sbr/adaptsearch commit 3c7982d775b6f3b472f6514d791edcb43cd258a1-dirty
abims-sbr
parents:
diff changeset
89 del filtered_bash_nuc[name]
eb95bf7f90ae planemo upload for repository https://github.com/abims-sbr/adaptsearch commit 3c7982d775b6f3b472f6514d791edcb43cd258a1-dirty
abims-sbr
parents:
diff changeset
90
eb95bf7f90ae planemo upload for repository https://github.com/abims-sbr/adaptsearch commit 3c7982d775b6f3b472f6514d791edcb43cd258a1-dirty
abims-sbr
parents:
diff changeset
91 return(filtered_bash_aa, filtered_bash_nuc)
eb95bf7f90ae planemo upload for repository https://github.com/abims-sbr/adaptsearch commit 3c7982d775b6f3b472f6514d791edcb43cd258a1-dirty
abims-sbr
parents:
diff changeset
92 ####################################
eb95bf7f90ae planemo upload for repository https://github.com/abims-sbr/adaptsearch commit 3c7982d775b6f3b472f6514d791edcb43cd258a1-dirty
abims-sbr
parents:
diff changeset
93
eb95bf7f90ae planemo upload for repository https://github.com/abims-sbr/adaptsearch commit 3c7982d775b6f3b472f6514d791edcb43cd258a1-dirty
abims-sbr
parents:
diff changeset
94
eb95bf7f90ae planemo upload for repository https://github.com/abims-sbr/adaptsearch commit 3c7982d775b6f3b472f6514d791edcb43cd258a1-dirty
abims-sbr
parents:
diff changeset
95 #######################
eb95bf7f90ae planemo upload for repository https://github.com/abims-sbr/adaptsearch commit 3c7982d775b6f3b472f6514d791edcb43cd258a1-dirty
abims-sbr
parents:
diff changeset
96 ##### RUN RUN RUN #####
eb95bf7f90ae planemo upload for repository https://github.com/abims-sbr/adaptsearch commit 3c7982d775b6f3b472f6514d791edcb43cd258a1-dirty
abims-sbr
parents:
diff changeset
97 #######################
eb95bf7f90ae planemo upload for repository https://github.com/abims-sbr/adaptsearch commit 3c7982d775b6f3b472f6514d791edcb43cd258a1-dirty
abims-sbr
parents:
diff changeset
98 import string, os, time, re, sys
eb95bf7f90ae planemo upload for repository https://github.com/abims-sbr/adaptsearch commit 3c7982d775b6f3b472f6514d791edcb43cd258a1-dirty
abims-sbr
parents:
diff changeset
99 from dico import dico
eb95bf7f90ae planemo upload for repository https://github.com/abims-sbr/adaptsearch commit 3c7982d775b6f3b472f6514d791edcb43cd258a1-dirty
abims-sbr
parents:
diff changeset
100
eb95bf7f90ae planemo upload for repository https://github.com/abims-sbr/adaptsearch commit 3c7982d775b6f3b472f6514d791edcb43cd258a1-dirty
abims-sbr
parents:
diff changeset
101 ### 0 ### PARAMETERS
eb95bf7f90ae planemo upload for repository https://github.com/abims-sbr/adaptsearch commit 3c7982d775b6f3b472f6514d791edcb43cd258a1-dirty
abims-sbr
parents:
diff changeset
102 MIN_SPECIES_NB = int(sys.argv[1])
eb95bf7f90ae planemo upload for repository https://github.com/abims-sbr/adaptsearch commit 3c7982d775b6f3b472f6514d791edcb43cd258a1-dirty
abims-sbr
parents:
diff changeset
103 MIN_LENGTH_FINAL_ALIGNMENT_NUC = int(sys.argv[2])
eb95bf7f90ae planemo upload for repository https://github.com/abims-sbr/adaptsearch commit 3c7982d775b6f3b472f6514d791edcb43cd258a1-dirty
abims-sbr
parents:
diff changeset
104 n0 = 0
eb95bf7f90ae planemo upload for repository https://github.com/abims-sbr/adaptsearch commit 3c7982d775b6f3b472f6514d791edcb43cd258a1-dirty
abims-sbr
parents:
diff changeset
105 bad = 0
eb95bf7f90ae planemo upload for repository https://github.com/abims-sbr/adaptsearch commit 3c7982d775b6f3b472f6514d791edcb43cd258a1-dirty
abims-sbr
parents:
diff changeset
106 good = 0
eb95bf7f90ae planemo upload for repository https://github.com/abims-sbr/adaptsearch commit 3c7982d775b6f3b472f6514d791edcb43cd258a1-dirty
abims-sbr
parents:
diff changeset
107 list_new_file = []
eb95bf7f90ae planemo upload for repository https://github.com/abims-sbr/adaptsearch commit 3c7982d775b6f3b472f6514d791edcb43cd258a1-dirty
abims-sbr
parents:
diff changeset
108 dicoco = {}
eb95bf7f90ae planemo upload for repository https://github.com/abims-sbr/adaptsearch commit 3c7982d775b6f3b472f6514d791edcb43cd258a1-dirty
abims-sbr
parents:
diff changeset
109 list_file = []
eb95bf7f90ae planemo upload for repository https://github.com/abims-sbr/adaptsearch commit 3c7982d775b6f3b472f6514d791edcb43cd258a1-dirty
abims-sbr
parents:
diff changeset
110 name_elems = ["orthogroup", "0", "with", "0", "species.fasta"]
eb95bf7f90ae planemo upload for repository https://github.com/abims-sbr/adaptsearch commit 3c7982d775b6f3b472f6514d791edcb43cd258a1-dirty
abims-sbr
parents:
diff changeset
111
eb95bf7f90ae planemo upload for repository https://github.com/abims-sbr/adaptsearch commit 3c7982d775b6f3b472f6514d791edcb43cd258a1-dirty
abims-sbr
parents:
diff changeset
112 ### 1 ### IN
eb95bf7f90ae planemo upload for repository https://github.com/abims-sbr/adaptsearch commit 3c7982d775b6f3b472f6514d791edcb43cd258a1-dirty
abims-sbr
parents:
diff changeset
113 path_IN1 = "./07_CDS_aa/"
eb95bf7f90ae planemo upload for repository https://github.com/abims-sbr/adaptsearch commit 3c7982d775b6f3b472f6514d791edcb43cd258a1-dirty
abims-sbr
parents:
diff changeset
114 L_IN1 = os.listdir(path_IN1)
eb95bf7f90ae planemo upload for repository https://github.com/abims-sbr/adaptsearch commit 3c7982d775b6f3b472f6514d791edcb43cd258a1-dirty
abims-sbr
parents:
diff changeset
115 lenght = len(L_IN1)
eb95bf7f90ae planemo upload for repository https://github.com/abims-sbr/adaptsearch commit 3c7982d775b6f3b472f6514d791edcb43cd258a1-dirty
abims-sbr
parents:
diff changeset
116 path_IN2 = "./07_CDS_nuc/"
eb95bf7f90ae planemo upload for repository https://github.com/abims-sbr/adaptsearch commit 3c7982d775b6f3b472f6514d791edcb43cd258a1-dirty
abims-sbr
parents:
diff changeset
117 L_IN2 = os.listdir(path_IN2)
eb95bf7f90ae planemo upload for repository https://github.com/abims-sbr/adaptsearch commit 3c7982d775b6f3b472f6514d791edcb43cd258a1-dirty
abims-sbr
parents:
diff changeset
118
eb95bf7f90ae planemo upload for repository https://github.com/abims-sbr/adaptsearch commit 3c7982d775b6f3b472f6514d791edcb43cd258a1-dirty
abims-sbr
parents:
diff changeset
119 ## 2 ## OUT
eb95bf7f90ae planemo upload for repository https://github.com/abims-sbr/adaptsearch commit 3c7982d775b6f3b472f6514d791edcb43cd258a1-dirty
abims-sbr
parents:
diff changeset
120 os.mkdir("08_CDS_aa_MINIMUM_MISSING_SEQUENCES")
eb95bf7f90ae planemo upload for repository https://github.com/abims-sbr/adaptsearch commit 3c7982d775b6f3b472f6514d791edcb43cd258a1-dirty
abims-sbr
parents:
diff changeset
121 path_OUT1 = "08_CDS_aa_MINIMUM_MISSING_SEQUENCES"
eb95bf7f90ae planemo upload for repository https://github.com/abims-sbr/adaptsearch commit 3c7982d775b6f3b472f6514d791edcb43cd258a1-dirty
abims-sbr
parents:
diff changeset
122 os.mkdir("08_CDS_nuc_MINIMUM_MISSING_SEQUENCES")
eb95bf7f90ae planemo upload for repository https://github.com/abims-sbr/adaptsearch commit 3c7982d775b6f3b472f6514d791edcb43cd258a1-dirty
abims-sbr
parents:
diff changeset
123 path_OUT2 = "08_CDS_nuc_MINIMUM_MISSING_SEQUENCES"
eb95bf7f90ae planemo upload for repository https://github.com/abims-sbr/adaptsearch commit 3c7982d775b6f3b472f6514d791edcb43cd258a1-dirty
abims-sbr
parents:
diff changeset
124
eb95bf7f90ae planemo upload for repository https://github.com/abims-sbr/adaptsearch commit 3c7982d775b6f3b472f6514d791edcb43cd258a1-dirty
abims-sbr
parents:
diff changeset
125
eb95bf7f90ae planemo upload for repository https://github.com/abims-sbr/adaptsearch commit 3c7982d775b6f3b472f6514d791edcb43cd258a1-dirty
abims-sbr
parents:
diff changeset
126 for file in L_IN1:
eb95bf7f90ae planemo upload for repository https://github.com/abims-sbr/adaptsearch commit 3c7982d775b6f3b472f6514d791edcb43cd258a1-dirty
abims-sbr
parents:
diff changeset
127 file_INaa = "%s/%s" %(path_IN1, file)
eb95bf7f90ae planemo upload for repository https://github.com/abims-sbr/adaptsearch commit 3c7982d775b6f3b472f6514d791edcb43cd258a1-dirty
abims-sbr
parents:
diff changeset
128 file_INnuc = "%s/%s" %(path_IN2, file)
eb95bf7f90ae planemo upload for repository https://github.com/abims-sbr/adaptsearch commit 3c7982d775b6f3b472f6514d791edcb43cd258a1-dirty
abims-sbr
parents:
diff changeset
129
eb95bf7f90ae planemo upload for repository https://github.com/abims-sbr/adaptsearch commit 3c7982d775b6f3b472f6514d791edcb43cd258a1-dirty
abims-sbr
parents:
diff changeset
130 dico_aa = dico(file_INaa) ### DEF 1 ###
eb95bf7f90ae planemo upload for repository https://github.com/abims-sbr/adaptsearch commit 3c7982d775b6f3b472f6514d791edcb43cd258a1-dirty
abims-sbr
parents:
diff changeset
131 dico_nuc = dico(file_INnuc) ### DEF 1 ###
eb95bf7f90ae planemo upload for repository https://github.com/abims-sbr/adaptsearch commit 3c7982d775b6f3b472f6514d791edcb43cd258a1-dirty
abims-sbr
parents:
diff changeset
132
eb95bf7f90ae planemo upload for repository https://github.com/abims-sbr/adaptsearch commit 3c7982d775b6f3b472f6514d791edcb43cd258a1-dirty
abims-sbr
parents:
diff changeset
133 if len(dico_aa) < MIN_SPECIES_NB :
eb95bf7f90ae planemo upload for repository https://github.com/abims-sbr/adaptsearch commit 3c7982d775b6f3b472f6514d791edcb43cd258a1-dirty
abims-sbr
parents:
diff changeset
134 list_file.append(file)
eb95bf7f90ae planemo upload for repository https://github.com/abims-sbr/adaptsearch commit 3c7982d775b6f3b472f6514d791edcb43cd258a1-dirty
abims-sbr
parents:
diff changeset
135
eb95bf7f90ae planemo upload for repository https://github.com/abims-sbr/adaptsearch commit 3c7982d775b6f3b472f6514d791edcb43cd258a1-dirty
abims-sbr
parents:
diff changeset
136 if list_file == lenght :
eb95bf7f90ae planemo upload for repository https://github.com/abims-sbr/adaptsearch commit 3c7982d775b6f3b472f6514d791edcb43cd258a1-dirty
abims-sbr
parents:
diff changeset
137 MIN_SPECIES_NB == MIN_SPECIES_NB - 1
eb95bf7f90ae planemo upload for repository https://github.com/abims-sbr/adaptsearch commit 3c7982d775b6f3b472f6514d791edcb43cd258a1-dirty
abims-sbr
parents:
diff changeset
138
eb95bf7f90ae planemo upload for repository https://github.com/abims-sbr/adaptsearch commit 3c7982d775b6f3b472f6514d791edcb43cd258a1-dirty
abims-sbr
parents:
diff changeset
139
eb95bf7f90ae planemo upload for repository https://github.com/abims-sbr/adaptsearch commit 3c7982d775b6f3b472f6514d791edcb43cd258a1-dirty
abims-sbr
parents:
diff changeset
140 for file in L_IN1 :
eb95bf7f90ae planemo upload for repository https://github.com/abims-sbr/adaptsearch commit 3c7982d775b6f3b472f6514d791edcb43cd258a1-dirty
abims-sbr
parents:
diff changeset
141 file_INaa = "%s/%s" %(path_IN1, file)
eb95bf7f90ae planemo upload for repository https://github.com/abims-sbr/adaptsearch commit 3c7982d775b6f3b472f6514d791edcb43cd258a1-dirty
abims-sbr
parents:
diff changeset
142 file_INnuc = "%s/%s" %(path_IN2, file)
eb95bf7f90ae planemo upload for repository https://github.com/abims-sbr/adaptsearch commit 3c7982d775b6f3b472f6514d791edcb43cd258a1-dirty
abims-sbr
parents:
diff changeset
143
eb95bf7f90ae planemo upload for repository https://github.com/abims-sbr/adaptsearch commit 3c7982d775b6f3b472f6514d791edcb43cd258a1-dirty
abims-sbr
parents:
diff changeset
144 dico_aa = dico(file_INaa) ### DEF 1 ###
eb95bf7f90ae planemo upload for repository https://github.com/abims-sbr/adaptsearch commit 3c7982d775b6f3b472f6514d791edcb43cd258a1-dirty
abims-sbr
parents:
diff changeset
145 dico_nuc = dico(file_INnuc) ### DEF 1 ###
eb95bf7f90ae planemo upload for repository https://github.com/abims-sbr/adaptsearch commit 3c7982d775b6f3b472f6514d791edcb43cd258a1-dirty
abims-sbr
parents:
diff changeset
146
eb95bf7f90ae planemo upload for repository https://github.com/abims-sbr/adaptsearch commit 3c7982d775b6f3b472f6514d791edcb43cd258a1-dirty
abims-sbr
parents:
diff changeset
147 ## 4.1 ## REMOVE POSITIONS WITH TOO MUCH MISSING DATA (i.e. not enough taxa represented at each position in the alignment)
eb95bf7f90ae planemo upload for repository https://github.com/abims-sbr/adaptsearch commit 3c7982d775b6f3b472f6514d791edcb43cd258a1-dirty
abims-sbr
parents:
diff changeset
148 filtered_bash_aa, filtered_bash_nuc = remove_position_with_too_much_missing_data(dico_aa, dico_nuc, MIN_SPECIES_NB) ### DEF 2 ###
eb95bf7f90ae planemo upload for repository https://github.com/abims-sbr/adaptsearch commit 3c7982d775b6f3b472f6514d791edcb43cd258a1-dirty
abims-sbr
parents:
diff changeset
149
eb95bf7f90ae planemo upload for repository https://github.com/abims-sbr/adaptsearch commit 3c7982d775b6f3b472f6514d791edcb43cd258a1-dirty
abims-sbr
parents:
diff changeset
150 k = filtered_bash_nuc.keys()
eb95bf7f90ae planemo upload for repository https://github.com/abims-sbr/adaptsearch commit 3c7982d775b6f3b472f6514d791edcb43cd258a1-dirty
abims-sbr
parents:
diff changeset
151 new_leng_nuc = 0
eb95bf7f90ae planemo upload for repository https://github.com/abims-sbr/adaptsearch commit 3c7982d775b6f3b472f6514d791edcb43cd258a1-dirty
abims-sbr
parents:
diff changeset
152 if k != []:
eb95bf7f90ae planemo upload for repository https://github.com/abims-sbr/adaptsearch commit 3c7982d775b6f3b472f6514d791edcb43cd258a1-dirty
abims-sbr
parents:
diff changeset
153 k0 = k[0]
eb95bf7f90ae planemo upload for repository https://github.com/abims-sbr/adaptsearch commit 3c7982d775b6f3b472f6514d791edcb43cd258a1-dirty
abims-sbr
parents:
diff changeset
154 seq0 = filtered_bash_nuc[k0]
eb95bf7f90ae planemo upload for repository https://github.com/abims-sbr/adaptsearch commit 3c7982d775b6f3b472f6514d791edcb43cd258a1-dirty
abims-sbr
parents:
diff changeset
155 new_leng_nuc = len(seq0)
eb95bf7f90ae planemo upload for repository https://github.com/abims-sbr/adaptsearch commit 3c7982d775b6f3b472f6514d791edcb43cd258a1-dirty
abims-sbr
parents:
diff changeset
156
eb95bf7f90ae planemo upload for repository https://github.com/abims-sbr/adaptsearch commit 3c7982d775b6f3b472f6514d791edcb43cd258a1-dirty
abims-sbr
parents:
diff changeset
157 ## 4.3 ## Change file name for output, depending the number of species remaining in the alignment
eb95bf7f90ae planemo upload for repository https://github.com/abims-sbr/adaptsearch commit 3c7982d775b6f3b472f6514d791edcb43cd258a1-dirty
abims-sbr
parents:
diff changeset
158 n0+=1
eb95bf7f90ae planemo upload for repository https://github.com/abims-sbr/adaptsearch commit 3c7982d775b6f3b472f6514d791edcb43cd258a1-dirty
abims-sbr
parents:
diff changeset
159 #name_elems[1] = str(n0)
eb95bf7f90ae planemo upload for repository https://github.com/abims-sbr/adaptsearch commit 3c7982d775b6f3b472f6514d791edcb43cd258a1-dirty
abims-sbr
parents:
diff changeset
160 name_elems[1] = file.split('_')[1]
eb95bf7f90ae planemo upload for repository https://github.com/abims-sbr/adaptsearch commit 3c7982d775b6f3b472f6514d791edcb43cd258a1-dirty
abims-sbr
parents:
diff changeset
161 name_elems[3] = str(len(filtered_bash_aa.keys()))
eb95bf7f90ae planemo upload for repository https://github.com/abims-sbr/adaptsearch commit 3c7982d775b6f3b472f6514d791edcb43cd258a1-dirty
abims-sbr
parents:
diff changeset
162 new_name = "_".join(name_elems)
eb95bf7f90ae planemo upload for repository https://github.com/abims-sbr/adaptsearch commit 3c7982d775b6f3b472f6514d791edcb43cd258a1-dirty
abims-sbr
parents:
diff changeset
163
eb95bf7f90ae planemo upload for repository https://github.com/abims-sbr/adaptsearch commit 3c7982d775b6f3b472f6514d791edcb43cd258a1-dirty
abims-sbr
parents:
diff changeset
164 ## 4.5 ## Write filtered alignment in OUTPUTs
eb95bf7f90ae planemo upload for repository https://github.com/abims-sbr/adaptsearch commit 3c7982d775b6f3b472f6514d791edcb43cd258a1-dirty
abims-sbr
parents:
diff changeset
165 ## aa
eb95bf7f90ae planemo upload for repository https://github.com/abims-sbr/adaptsearch commit 3c7982d775b6f3b472f6514d791edcb43cd258a1-dirty
abims-sbr
parents:
diff changeset
166 if filtered_bash_aa != {} and new_leng_nuc >= MIN_LENGTH_FINAL_ALIGNMENT_NUC:
eb95bf7f90ae planemo upload for repository https://github.com/abims-sbr/adaptsearch commit 3c7982d775b6f3b472f6514d791edcb43cd258a1-dirty
abims-sbr
parents:
diff changeset
167 OUTaa=open("%s/%s" %(path_OUT1, new_name), "w")
eb95bf7f90ae planemo upload for repository https://github.com/abims-sbr/adaptsearch commit 3c7982d775b6f3b472f6514d791edcb43cd258a1-dirty
abims-sbr
parents:
diff changeset
168 for fasta_name in filtered_bash_aa.keys():
eb95bf7f90ae planemo upload for repository https://github.com/abims-sbr/adaptsearch commit 3c7982d775b6f3b472f6514d791edcb43cd258a1-dirty
abims-sbr
parents:
diff changeset
169 seq_aa = filtered_bash_aa[fasta_name]
eb95bf7f90ae planemo upload for repository https://github.com/abims-sbr/adaptsearch commit 3c7982d775b6f3b472f6514d791edcb43cd258a1-dirty
abims-sbr
parents:
diff changeset
170 OUTaa.write("%s\n" %fasta_name)
eb95bf7f90ae planemo upload for repository https://github.com/abims-sbr/adaptsearch commit 3c7982d775b6f3b472f6514d791edcb43cd258a1-dirty
abims-sbr
parents:
diff changeset
171 OUTaa.write("%s\n" %seq_aa)
eb95bf7f90ae planemo upload for repository https://github.com/abims-sbr/adaptsearch commit 3c7982d775b6f3b472f6514d791edcb43cd258a1-dirty
abims-sbr
parents:
diff changeset
172 OUTaa.close()
eb95bf7f90ae planemo upload for repository https://github.com/abims-sbr/adaptsearch commit 3c7982d775b6f3b472f6514d791edcb43cd258a1-dirty
abims-sbr
parents:
diff changeset
173 # nuc
eb95bf7f90ae planemo upload for repository https://github.com/abims-sbr/adaptsearch commit 3c7982d775b6f3b472f6514d791edcb43cd258a1-dirty
abims-sbr
parents:
diff changeset
174 if filtered_bash_nuc != {} and new_leng_nuc >= MIN_LENGTH_FINAL_ALIGNMENT_NUC:
eb95bf7f90ae planemo upload for repository https://github.com/abims-sbr/adaptsearch commit 3c7982d775b6f3b472f6514d791edcb43cd258a1-dirty
abims-sbr
parents:
diff changeset
175 good+=1
eb95bf7f90ae planemo upload for repository https://github.com/abims-sbr/adaptsearch commit 3c7982d775b6f3b472f6514d791edcb43cd258a1-dirty
abims-sbr
parents:
diff changeset
176 OUTnuc=open("%s/%s" %(path_OUT2, new_name), "w")
eb95bf7f90ae planemo upload for repository https://github.com/abims-sbr/adaptsearch commit 3c7982d775b6f3b472f6514d791edcb43cd258a1-dirty
abims-sbr
parents:
diff changeset
177 for fasta_name in filtered_bash_nuc.keys():
eb95bf7f90ae planemo upload for repository https://github.com/abims-sbr/adaptsearch commit 3c7982d775b6f3b472f6514d791edcb43cd258a1-dirty
abims-sbr
parents:
diff changeset
178 seq_nuc = filtered_bash_nuc[fasta_name]
eb95bf7f90ae planemo upload for repository https://github.com/abims-sbr/adaptsearch commit 3c7982d775b6f3b472f6514d791edcb43cd258a1-dirty
abims-sbr
parents:
diff changeset
179 OUTnuc.write("%s\n" %fasta_name)
eb95bf7f90ae planemo upload for repository https://github.com/abims-sbr/adaptsearch commit 3c7982d775b6f3b472f6514d791edcb43cd258a1-dirty
abims-sbr
parents:
diff changeset
180 OUTnuc.write("%s\n" %seq_nuc)
eb95bf7f90ae planemo upload for repository https://github.com/abims-sbr/adaptsearch commit 3c7982d775b6f3b472f6514d791edcb43cd258a1-dirty
abims-sbr
parents:
diff changeset
181 OUTnuc.close()
eb95bf7f90ae planemo upload for repository https://github.com/abims-sbr/adaptsearch commit 3c7982d775b6f3b472f6514d791edcb43cd258a1-dirty
abims-sbr
parents:
diff changeset
182 else:
eb95bf7f90ae planemo upload for repository https://github.com/abims-sbr/adaptsearch commit 3c7982d775b6f3b472f6514d791edcb43cd258a1-dirty
abims-sbr
parents:
diff changeset
183 bad+=1
eb95bf7f90ae planemo upload for repository https://github.com/abims-sbr/adaptsearch commit 3c7982d775b6f3b472f6514d791edcb43cd258a1-dirty
abims-sbr
parents:
diff changeset
184
eb95bf7f90ae planemo upload for repository https://github.com/abims-sbr/adaptsearch commit 3c7982d775b6f3b472f6514d791edcb43cd258a1-dirty
abims-sbr
parents:
diff changeset
185
eb95bf7f90ae planemo upload for repository https://github.com/abims-sbr/adaptsearch commit 3c7982d775b6f3b472f6514d791edcb43cd258a1-dirty
abims-sbr
parents:
diff changeset
186 ## 5 ## Print
eb95bf7f90ae planemo upload for repository https://github.com/abims-sbr/adaptsearch commit 3c7982d775b6f3b472f6514d791edcb43cd258a1-dirty
abims-sbr
parents:
diff changeset
187 print "*************** 2nd Filter : removal of the indel ***************"
eb95bf7f90ae planemo upload for repository https://github.com/abims-sbr/adaptsearch commit 3c7982d775b6f3b472f6514d791edcb43cd258a1-dirty
abims-sbr
parents:
diff changeset
188 print "\nTotal number of locus recorded = %d" %n0
eb95bf7f90ae planemo upload for repository https://github.com/abims-sbr/adaptsearch commit 3c7982d775b6f3b472f6514d791edcb43cd258a1-dirty
abims-sbr
parents:
diff changeset
189 print "\tTotal number of locus with no indels (SAVED) = %d" %good
eb95bf7f90ae planemo upload for repository https://github.com/abims-sbr/adaptsearch commit 3c7982d775b6f3b472f6514d791edcb43cd258a1-dirty
abims-sbr
parents:
diff changeset
190 print "\tTotal number of locus, when removing indel, wich are empty (EXCLUDED) = %d" %bad
eb95bf7f90ae planemo upload for repository https://github.com/abims-sbr/adaptsearch commit 3c7982d775b6f3b472f6514d791edcb43cd258a1-dirty
abims-sbr
parents:
diff changeset
191 print ""