Mercurial > repos > abims-sbr > concatphyl

annotate scripts/S01_concatenate.py @ 10:5cae58aa1181 draft default tip

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 3c7982d775b6f3b472f6514d791edcb43cd258a1
author lecorguille
date Mon, 24 Sep 2018 03:59:21 -0400
parents ff1a3a790363
children
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
rev   line source
1
f181bd945a6c planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
1 #!/usr/bin/python
f181bd945a6c planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
2 ## Author: Eric Fontanillas
f181bd945a6c planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
3 ## Last modification: 17/06/2011
f181bd945a6c planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
4 ## Subject: find and remove indels
f181bd945a6c planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
5
f181bd945a6c planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
6 ###############################
f181bd945a6c planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
7 ##### DEF 0 : Dico fasta #####
f181bd945a6c planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
8 ###############################
f181bd945a6c planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
9 def dico(F2):
f181bd945a6c planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
10 dicoco = {}
7
59f4b9ffd92b planemo upload for repository https://github.com/abims-sbr/adaptsearch commit b7a3030ea134b5dfad89b1a869db659d72d1145c
abims-sbr
parents: 2
diff changeset
11 with open(F2, "r") as file:
59f4b9ffd92b planemo upload for repository https://github.com/abims-sbr/adaptsearch commit b7a3030ea134b5dfad89b1a869db659d72d1145c
abims-sbr
parents: 2
diff changeset
12 for name, query in itertools.izip_longest(*[file]*2):
59f4b9ffd92b planemo upload for repository https://github.com/abims-sbr/adaptsearch commit b7a3030ea134b5dfad89b1a869db659d72d1145c
abims-sbr
parents: 2
diff changeset
13 if not name:
59f4b9ffd92b planemo upload for repository https://github.com/abims-sbr/adaptsearch commit b7a3030ea134b5dfad89b1a869db659d72d1145c
abims-sbr
parents: 2
diff changeset
14 break
59f4b9ffd92b planemo upload for repository https://github.com/abims-sbr/adaptsearch commit b7a3030ea134b5dfad89b1a869db659d72d1145c
abims-sbr
parents: 2
diff changeset
15 if name[0] == ">":
59f4b9ffd92b planemo upload for repository https://github.com/abims-sbr/adaptsearch commit b7a3030ea134b5dfad89b1a869db659d72d1145c
abims-sbr
parents: 2
diff changeset
16 fasta_name_query = name[:-1]
59f4b9ffd92b planemo upload for repository https://github.com/abims-sbr/adaptsearch commit b7a3030ea134b5dfad89b1a869db659d72d1145c
abims-sbr
parents: 2
diff changeset
17 Sn = string.split(fasta_name_query, "||")
59f4b9ffd92b planemo upload for repository https://github.com/abims-sbr/adaptsearch commit b7a3030ea134b5dfad89b1a869db659d72d1145c
abims-sbr
parents: 2
diff changeset
18 fasta_name_query = Sn[0]
59f4b9ffd92b planemo upload for repository https://github.com/abims-sbr/adaptsearch commit b7a3030ea134b5dfad89b1a869db659d72d1145c
abims-sbr
parents: 2
diff changeset
19 fasta_seq_query = query[:-1]
59f4b9ffd92b planemo upload for repository https://github.com/abims-sbr/adaptsearch commit b7a3030ea134b5dfad89b1a869db659d72d1145c
abims-sbr
parents: 2
diff changeset
20 dicoco[fasta_name_query]=fasta_seq_query
59f4b9ffd92b planemo upload for repository https://github.com/abims-sbr/adaptsearch commit b7a3030ea134b5dfad89b1a869db659d72d1145c
abims-sbr
parents: 2
diff changeset
21 return dicoco
1
f181bd945a6c planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
22 ###################################################################################
f181bd945a6c planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
23
f181bd945a6c planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
24
f181bd945a6c planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
25 ####################
f181bd945a6c planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
26 ###### DEF 11 ######
f181bd945a6c planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
27 ####################
f181bd945a6c planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
28 ## Concatenate sequences
f181bd945a6c planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
29 ###########################
2
1f8d039bd241 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit 44a89d5eeb82789bfc643b33c11f391281b6374b
abims-sbr
parents: 1
diff changeset
30 def concatenate(L_IN, SPECIES_ID_LIST):
1
f181bd945a6c planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
31 ## 4 ## Process files
f181bd945a6c planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
32 ## 4.1 ## Create the bash and the fasta names entries (name of the species)
f181bd945a6c planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
33 bash_concat = {}
2
1f8d039bd241 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit 44a89d5eeb82789bfc643b33c11f391281b6374b
abims-sbr
parents: 1
diff changeset
34
1
f181bd945a6c planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
35 for species_ID in SPECIES_ID_LIST:
f181bd945a6c planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
36 bash_concat[species_ID] = ''
f181bd945a6c planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
37
f181bd945a6c planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
38 ln_concat = 0
f181bd945a6c planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
39 nb_locus = 0
f181bd945a6c planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
40 pos=1
f181bd945a6c planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
41 list_genes_position=[]
f181bd945a6c planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
42 ## 4.2 ## Concatenate
f181bd945a6c planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
43 for file in L_IN:
f181bd945a6c planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
44 nb_locus=nb_locus+1
2
1f8d039bd241 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit 44a89d5eeb82789bfc643b33c11f391281b6374b
abims-sbr
parents: 1
diff changeset
45
7
59f4b9ffd92b planemo upload for repository https://github.com/abims-sbr/adaptsearch commit b7a3030ea134b5dfad89b1a869db659d72d1145c
abims-sbr
parents: 2
diff changeset
46 ## a ## Open alignments
59f4b9ffd92b planemo upload for repository https://github.com/abims-sbr/adaptsearch commit b7a3030ea134b5dfad89b1a869db659d72d1145c
abims-sbr
parents: 2
diff changeset
47 dico_seq = dico(file) ### DEF 0 ###
1
f181bd945a6c planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
48 ## b ## Get alignment length + genes positions for RAxML
f181bd945a6c planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
49 key0 = dico_seq.keys()[0]
f181bd945a6c planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
50 ln = len(dico_seq[key0])
f181bd945a6c planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
51 ln_concat = ln_concat + ln
f181bd945a6c planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
52
f181bd945a6c planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
53 pos_start = pos
f181bd945a6c planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
54 pos_end = pos+ln-1
f181bd945a6c planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
55 pos=pos_end+1
f181bd945a6c planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
56 position="%d-%d" %(pos_start, pos_end)
f181bd945a6c planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
57 RAxML_name = file[:-6]
f181bd945a6c planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
58 sublist = [RAxML_name, position]
f181bd945a6c planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
59 list_genes_position.append(sublist)
2
1f8d039bd241 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit 44a89d5eeb82789bfc643b33c11f391281b6374b
abims-sbr
parents: 1
diff changeset
60
1
f181bd945a6c planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
61 ## c ## Generate "empty" sequence with alignment length * "-"
f181bd945a6c planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
62 empty_seq = "-" * ln
2
1f8d039bd241 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit 44a89d5eeb82789bfc643b33c11f391281b6374b
abims-sbr
parents: 1
diff changeset
63
1
f181bd945a6c planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
64 ## d ## Concatenate
f181bd945a6c planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
65 ## d.1 ## Detect missing species in this alignment
f181bd945a6c planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
66 list_ID=[]
f181bd945a6c planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
67 list_absent_ID=[]
f181bd945a6c planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
68 bash_fastaName={}
f181bd945a6c planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
69 for fasta_name in dico_seq:
f181bd945a6c planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
70 ID = fasta_name[1:3]
f181bd945a6c planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
71 list_ID.append(ID)
f181bd945a6c planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
72 seq = dico_seq[fasta_name]
f181bd945a6c planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
73 bash_fastaName[ID]=fasta_name
f181bd945a6c planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
74 for sp_ID in SPECIES_ID_LIST:
f181bd945a6c planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
75 if sp_ID not in list_ID:
f181bd945a6c planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
76 list_absent_ID.append(sp_ID)
f181bd945a6c planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
77
f181bd945a6c planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
78 for ID in SPECIES_ID_LIST:
f181bd945a6c planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
79 if ID in list_absent_ID:
f181bd945a6c planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
80 bash_concat[ID] = bash_concat[ID] + empty_seq
f181bd945a6c planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
81 else:
f181bd945a6c planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
82 fasta_name = bash_fastaName[ID]
f181bd945a6c planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
83 seq = dico_seq[fasta_name]
2
1f8d039bd241 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit 44a89d5eeb82789bfc643b33c11f391281b6374b
abims-sbr
parents: 1
diff changeset
84 bash_concat[ID] = bash_concat[ID] + seq
1
f181bd945a6c planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
85
f181bd945a6c planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
86 return(bash_concat, ln_concat, nb_locus, list_genes_position)
f181bd945a6c planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
87 ####################################
f181bd945a6c planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
88
f181bd945a6c planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
89
f181bd945a6c planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
90 ########################################
f181bd945a6c planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
91 ##### DEF 12 : get codon position #####
f181bd945a6c planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
92 ########################################
f181bd945a6c planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
93 def get_codon_position(seq_inORF):
f181bd945a6c planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
94
f181bd945a6c planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
95 ln = len(seq_inORF)
f181bd945a6c planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
96
f181bd945a6c planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
97 i=0
f181bd945a6c planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
98 seq_pos1=""
f181bd945a6c planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
99 seq_pos2=""
f181bd945a6c planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
100 seq_pos12=""
f181bd945a6c planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
101 seq_pos3=""
f181bd945a6c planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
102 while i<ln:
f181bd945a6c planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
103 pos1 = seq_inORF[i]
f181bd945a6c planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
104 pos2 = seq_inORF[i+1]
f181bd945a6c planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
105 pos3 = seq_inORF[i+2]
2
1f8d039bd241 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit 44a89d5eeb82789bfc643b33c11f391281b6374b
abims-sbr
parents: 1
diff changeset
106
1
f181bd945a6c planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
107 seq_pos1 = seq_pos1 + pos1
f181bd945a6c planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
108 seq_pos2 = seq_pos2 + pos2
f181bd945a6c planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
109 seq_pos12 = seq_pos12 + pos1 + pos2
f181bd945a6c planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
110 seq_pos3 = seq_pos3 + pos3
f181bd945a6c planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
111
f181bd945a6c planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
112 i = i+3
f181bd945a6c planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
113
f181bd945a6c planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
114 return(seq_pos1, seq_pos2, seq_pos12, seq_pos3)
f181bd945a6c planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
115 ###############################################################################
f181bd945a6c planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
116
f181bd945a6c planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
117
f181bd945a6c planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
118
f181bd945a6c planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
119 #######################
f181bd945a6c planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
120 ##### RUN RUN RUN #####
f181bd945a6c planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
121 #######################
7
59f4b9ffd92b planemo upload for repository https://github.com/abims-sbr/adaptsearch commit b7a3030ea134b5dfad89b1a869db659d72d1145c
abims-sbr
parents: 2
diff changeset
122 import string, os, time, re, sys, itertools
1
f181bd945a6c planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
123
f181bd945a6c planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
124 list_species = []
f181bd945a6c planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
125 SPECIES_ID_LIST = []
f181bd945a6c planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
126 fasta = "^.*fasta$"
f181bd945a6c planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
127 i=3
f181bd945a6c planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
128
2
1f8d039bd241 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit 44a89d5eeb82789bfc643b33c11f391281b6374b
abims-sbr
parents: 1
diff changeset
129 ## Arguments
1f8d039bd241 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit 44a89d5eeb82789bfc643b33c11f391281b6374b
abims-sbr
parents: 1
diff changeset
130 infiles_filter_assemblies = sys.argv[1]
1f8d039bd241 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit 44a89d5eeb82789bfc643b33c11f391281b6374b
abims-sbr
parents: 1
diff changeset
131 format_run = sys.argv[2]
1f8d039bd241 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit 44a89d5eeb82789bfc643b33c11f391281b6374b
abims-sbr
parents: 1
diff changeset
132
1
f181bd945a6c planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
133 ## add file to list_species
2
1f8d039bd241 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit 44a89d5eeb82789bfc643b33c11f391281b6374b
abims-sbr
parents: 1
diff changeset
134 list_species = str.split(infiles_filter_assemblies,",")
1
f181bd945a6c planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
135
f181bd945a6c planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
136 ## in SPECIES_ID_LIST, only the 2 first letters of name of species
f181bd945a6c planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
137 for name in list_species :
f181bd945a6c planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
138 name = name[:2]
f181bd945a6c planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
139 SPECIES_ID_LIST.append(name)
f181bd945a6c planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
140
2
1f8d039bd241 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit 44a89d5eeb82789bfc643b33c11f391281b6374b
abims-sbr
parents: 1
diff changeset
141 ## add alignment files to L_IN
9
ff1a3a790363 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit f1ba8d136e0129f3e8435b25a95f70f697d51464-dirty
abims-sbr
parents: 7
diff changeset
142 list_files = []
ff1a3a790363 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit f1ba8d136e0129f3e8435b25a95f70f697d51464-dirty
abims-sbr
parents: 7
diff changeset
143 with open(sys.argv[3], 'r') as f:
ff1a3a790363 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit f1ba8d136e0129f3e8435b25a95f70f697d51464-dirty
abims-sbr
parents: 7
diff changeset
144 for line in f.readlines():
ff1a3a790363 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit f1ba8d136e0129f3e8435b25a95f70f697d51464-dirty
abims-sbr
parents: 7
diff changeset
145 list_files.append(line.strip('\n'))
ff1a3a790363 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit f1ba8d136e0129f3e8435b25a95f70f697d51464-dirty
abims-sbr
parents: 7
diff changeset
146
ff1a3a790363 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit f1ba8d136e0129f3e8435b25a95f70f697d51464-dirty
abims-sbr
parents: 7
diff changeset
147 L_IN = []
ff1a3a790363 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit f1ba8d136e0129f3e8435b25a95f70f697d51464-dirty
abims-sbr
parents: 7
diff changeset
148 for file in list_files:
ff1a3a790363 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit f1ba8d136e0129f3e8435b25a95f70f697d51464-dirty
abims-sbr
parents: 7
diff changeset
149 L_IN.append(file)
ff1a3a790363 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit f1ba8d136e0129f3e8435b25a95f70f697d51464-dirty
abims-sbr
parents: 7
diff changeset
150
ff1a3a790363 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit f1ba8d136e0129f3e8435b25a95f70f697d51464-dirty
abims-sbr
parents: 7
diff changeset
151 #L_IN = str.split(input_alignments,",")
2
1f8d039bd241 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit 44a89d5eeb82789bfc643b33c11f391281b6374b
abims-sbr
parents: 1
diff changeset
152 print(L_IN)
1
f181bd945a6c planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
153
f181bd945a6c planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
154 ### 1 ### Proteic
2
1f8d039bd241 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit 44a89d5eeb82789bfc643b33c11f391281b6374b
abims-sbr
parents: 1
diff changeset
155 if format_run == "proteic" :
1
f181bd945a6c planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
156
f181bd945a6c planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
157 OUT1 = open("02_Concatenation_aa.fas", "w")
f181bd945a6c planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
158 OUT2 = open("02_Concatenation_aa.phy", "w")
f181bd945a6c planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
159 OUT3 = open("02_Concatenation_aa.nex", "w")
f181bd945a6c planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
160 OUT_PARTITION_gene_AA = open("06_partitions_gene_AA","w")
f181bd945a6c planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
161
f181bd945a6c planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
162
f181bd945a6c planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
163 ## Get bash with concatenation
2
1f8d039bd241 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit 44a89d5eeb82789bfc643b33c11f391281b6374b
abims-sbr
parents: 1
diff changeset
164 bash_concatenation, ln, nb_locus,list_genes_position= concatenate(L_IN, SPECIES_ID_LIST) ### DEF 11 ##
1
f181bd945a6c planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
165
f181bd945a6c planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
166 ## Write gene AA partition file for RAxML
f181bd945a6c planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
167 for sublist in list_genes_position:
f181bd945a6c planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
168 name = sublist[0]
f181bd945a6c planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
169 positions=sublist[1]
f181bd945a6c planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
170 OUT_PARTITION_gene_AA.write("DNA,%s=%s\n"%(name,positions))
f181bd945a6c planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
171 OUT_PARTITION_gene_AA.close()
f181bd945a6c planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
172
f181bd945a6c planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
173 ## Get "ntax" for NEXUS HEADER
f181bd945a6c planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
174 nb_taxa = len(bash_concatenation.keys())
f181bd945a6c planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
175
f181bd945a6c planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
176 print "******************** CONCATENATION ********************\n"
f181bd945a6c planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
177 print "Process amino-acid concatenation:"
f181bd945a6c planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
178 print "\tNumber of taxa aligned = %d" %nb_taxa
f181bd945a6c planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
179 print "\tNumber of loci concatenated = %d\n" %nb_locus
f181bd945a6c planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
180 print "\tTotal length of the concatenated sequences = %d" %ln
f181bd945a6c planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
181
f181bd945a6c planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
182
f181bd945a6c planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
183 ## Print NEXUS HEADER:
f181bd945a6c planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
184 OUT3.write("#NEXUS\n\n")
f181bd945a6c planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
185 OUT3.write("Begin data;\n")
f181bd945a6c planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
186 OUT3.write("\tDimensions ntax=%d nchar=%d;\n" %(nb_taxa, ln))
f181bd945a6c planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
187 OUT3.write("\tFormat datatype=aa gap=-;\n")
f181bd945a6c planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
188 OUT3.write("\tMatrix\n")
f181bd945a6c planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
189
f181bd945a6c planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
190 ## Print PHYLIP HEADER:
f181bd945a6c planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
191 OUT2.write(" %d %d\n" %(nb_taxa, ln))
f181bd945a6c planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
192
f181bd945a6c planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
193 ## 3.5 ## Print outputs
f181bd945a6c planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
194 for seq_name in bash_concatenation.keys():
f181bd945a6c planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
195 seq = bash_concatenation[seq_name]
2
1f8d039bd241 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit 44a89d5eeb82789bfc643b33c11f391281b6374b
abims-sbr
parents: 1
diff changeset
196
1
f181bd945a6c planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
197 ## Filtering the sequence in case of remaining "?"
2
1f8d039bd241 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit 44a89d5eeb82789bfc643b33c11f391281b6374b
abims-sbr
parents: 1
diff changeset
198 seq = string.replace(seq, "?", "-")
1
f181bd945a6c planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
199
f181bd945a6c planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
200 #print seq FASTA FORMAT
f181bd945a6c planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
201 OUT1.write(">%s\n" %seq_name)
f181bd945a6c planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
202 OUT1.write("%s\n" %seq)
2
1f8d039bd241 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit 44a89d5eeb82789bfc643b33c11f391281b6374b
abims-sbr
parents: 1
diff changeset
203
1
f181bd945a6c planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
204 #print seq PHYLIP FORMAT
f181bd945a6c planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
205 OUT2.write("%s\n" %seq_name)
f181bd945a6c planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
206 OUT2.write("%s\n" %seq)
f181bd945a6c planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
207
f181bd945a6c planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
208 #print seq NEXUS FORMAT
f181bd945a6c planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
209 OUT3.write("%s" %seq_name)
f181bd945a6c planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
210 OUT3.write(" %s\n" %seq)
f181bd945a6c planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
211
f181bd945a6c planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
212 OUT3.write("\t;\n")
f181bd945a6c planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
213 OUT3.write("End;\n")
f181bd945a6c planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
214 OUT1.close()
f181bd945a6c planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
215 OUT2.close()
f181bd945a6c planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
216 OUT2.close()
f181bd945a6c planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
217
f181bd945a6c planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
218
f181bd945a6c planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
219 ### 2 ### Nucleic
2
1f8d039bd241 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit 44a89d5eeb82789bfc643b33c11f391281b6374b
abims-sbr
parents: 1
diff changeset
220 elif format_run == "nucleic" :
1
f181bd945a6c planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
221
f181bd945a6c planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
222 OUT1 = open("03_Concatenation_nuc.fas", "w")
f181bd945a6c planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
223 OUT2 = open("03_Concatenation_nuc.phy", "w")
f181bd945a6c planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
224 OUT3 = open("03_Concatenation_nuc.nex", "w")
f181bd945a6c planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
225
f181bd945a6c planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
226 OUT1_pos12 = open("03_Concatenation_pos12_nuc.fas", "w")
f181bd945a6c planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
227 OUT2_pos12 = open("03_Concatenation_pos12_nuc.phy", "w")
f181bd945a6c planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
228 OUT3_pos12 = open("03_Concatenation_pos12_nuc.nex", "w")
f181bd945a6c planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
229
f181bd945a6c planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
230 OUT1_pos3 = open("03_Concatenation_pos3_nuc.fas", "w")
f181bd945a6c planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
231 OUT2_pos3 = open("03_Concatenation_pos3_nuc.phy", "w")
f181bd945a6c planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
232 OUT3_pos3 = open("03_Concatenation_pos3_nuc.nex", "w")
f181bd945a6c planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
233
f181bd945a6c planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
234 OUT_PARTITION_codon_12_3 = open("05_partitions_codon12_3","w")
f181bd945a6c planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
235 OUT_PARTITION_gene_NUC = open("05_partitions_gene_NUC","w")
f181bd945a6c planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
236 OUT_PARTITION_gene_PLUS_codon_12_3 = open("05_partitions_gene_PLUS_codon12_3","w")
f181bd945a6c planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
237
f181bd945a6c planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
238 ## Get bash with concatenation
2
1f8d039bd241 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit 44a89d5eeb82789bfc643b33c11f391281b6374b
abims-sbr
parents: 1
diff changeset
239 bash_concatenation, ln, nb_locus, list_genes_position = concatenate(L_IN, SPECIES_ID_LIST) ### DEF 11 ##
1
f181bd945a6c planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
240 ln_12 = ln/3*2 ### length of the alignment when only the 2 first codon position
f181bd945a6c planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
241 ln_3 = ln/3 ### length of the alignment when only the third codon position
f181bd945a6c planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
242
f181bd945a6c planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
243 ## Write partition files for RAxML subsequent runs
f181bd945a6c planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
244 # a # Codon partition
f181bd945a6c planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
245 OUT_PARTITION_codon_12_3.write("DNA, p1=1-%d\\3,2-%d\\3\n" %(ln, ln))
f181bd945a6c planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
246 OUT_PARTITION_codon_12_3.write("DNA, p2=3-%d\\3\n" %(ln))
f181bd945a6c planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
247 OUT_PARTITION_codon_12_3.close()
f181bd945a6c planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
248
f181bd945a6c planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
249 # b # Gene partition
f181bd945a6c planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
250 for sublist in list_genes_position:
f181bd945a6c planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
251 name=sublist[0]
f181bd945a6c planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
252 positions=sublist[1]
f181bd945a6c planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
253 OUT_PARTITION_gene_NUC.write("DNA,%s=%s\n"%(name,positions))
f181bd945a6c planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
254 OUT_PARTITION_gene_NUC.close()
f181bd945a6c planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
255
f181bd945a6c planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
256 # c # Mixed partition (codon + gene)
f181bd945a6c planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
257 for sublist in list_genes_position:
f181bd945a6c planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
258 name = sublist[0]
f181bd945a6c planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
259 positions = sublist[1]
f181bd945a6c planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
260 S1 = string.split(positions, "-")
f181bd945a6c planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
261 pos_start1 = string.atoi(S1[0])
f181bd945a6c planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
262 pos_end = string.atoi(S1[1])
f181bd945a6c planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
263 pos_start2=pos_start1+1
f181bd945a6c planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
264 pos_start3=pos_start2+1
f181bd945a6c planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
265 partition1 = "DNA, %s_1=%d-%d\\3,%d-%d\\3\n" %(name,pos_start1, pos_end, pos_start2, pos_end)
f181bd945a6c planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
266 partition2 = "DNA, %s_2=%d-%d\\3\n" %(name,pos_start3, pos_end)
f181bd945a6c planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
267 OUT_PARTITION_gene_PLUS_codon_12_3.write(partition1)
f181bd945a6c planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
268 OUT_PARTITION_gene_PLUS_codon_12_3.write(partition2)
2
1f8d039bd241 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit 44a89d5eeb82789bfc643b33c11f391281b6374b
abims-sbr
parents: 1
diff changeset
269
1
f181bd945a6c planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
270 OUT_PARTITION_gene_PLUS_codon_12_3.close()
f181bd945a6c planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
271
f181bd945a6c planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
272
f181bd945a6c planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
273 ## Get "ntax" for NEXUS HEADER
f181bd945a6c planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
274 nb_taxa = len(bash_concatenation.keys())
f181bd945a6c planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
275
f181bd945a6c planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
276 print "******************** CONCATENATION ********************\n"
f181bd945a6c planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
277 print "Process nucleotides concatenation:"
f181bd945a6c planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
278 print "\tNumber of taxa aligned = %d" %nb_taxa
f181bd945a6c planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
279 print "\tNumber of loci concatenated = %d\n" %nb_locus
f181bd945a6c planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
280 print "\tTotal length of the concatenated sequences [All codon positions] = %d" %ln
f181bd945a6c planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
281 print "\t\tTotal length of the concatenated sequences [Codon positions 1 & 2] = %d" %ln_12
f181bd945a6c planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
282 print "\t\tTotal length of the concatenated sequences [Codon position 3] = %d" %ln_3
f181bd945a6c planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
283
f181bd945a6c planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
284
f181bd945a6c planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
285 ## Print NEXUS HEADER:
f181bd945a6c planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
286 OUT3.write("#NEXUS\n\n")
f181bd945a6c planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
287 OUT3.write("Begin data;\n")
f181bd945a6c planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
288 OUT3.write("\tDimensions ntax=%d nchar=%d;\n" %(nb_taxa, ln))
f181bd945a6c planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
289 OUT3.write("\tFormat datatype=dna gap=-;\n")
f181bd945a6c planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
290 OUT3.write("\tMatrix\n")
f181bd945a6c planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
291
f181bd945a6c planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
292 OUT3_pos12.write("#NEXUS\n\n")
f181bd945a6c planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
293 OUT3_pos12.write("Begin data;\n")
f181bd945a6c planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
294 OUT3_pos12.write("\tDimensions ntax=%d nchar=%d;\n" %(nb_taxa, ln_12))
f181bd945a6c planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
295 OUT3_pos12.write("\tFormat datatype=dna gap=-;\n")
f181bd945a6c planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
296 OUT3_pos12.write("\tMatrix\n")
f181bd945a6c planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
297
f181bd945a6c planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
298 OUT3_pos3.write("#NEXUS\n\n")
f181bd945a6c planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
299 OUT3_pos3.write("Begin data;\n")
f181bd945a6c planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
300 OUT3_pos3.write("\tDimensions ntax=%d nchar=%d;\n" %(nb_taxa, ln_3))
f181bd945a6c planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
301 OUT3_pos3.write("\tFormat datatype=dna gap=-;\n")
f181bd945a6c planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
302 OUT3_pos3.write("\tMatrix\n")
f181bd945a6c planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
303
f181bd945a6c planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
304 ## Print PHYLIP HEADER:
f181bd945a6c planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
305 OUT2.write(" %d %d\n" %(nb_taxa, ln))
f181bd945a6c planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
306 OUT2_pos12.write(" %d %d\n" %(nb_taxa, ln_12))
f181bd945a6c planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
307 OUT2_pos3.write(" %d %d\n" %(nb_taxa, ln_3))
f181bd945a6c planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
308
f181bd945a6c planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
309 ## Print outputs
f181bd945a6c planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
310 for seq_name in bash_concatenation.keys():
f181bd945a6c planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
311 seq = bash_concatenation[seq_name]
2
1f8d039bd241 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit 44a89d5eeb82789bfc643b33c11f391281b6374b
abims-sbr
parents: 1
diff changeset
312
1
f181bd945a6c planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
313 ## Filtering the sequence in case of remaining "?"
2
1f8d039bd241 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit 44a89d5eeb82789bfc643b33c11f391281b6374b
abims-sbr
parents: 1
diff changeset
314 seq = string.replace(seq, "?", "-")
1
f181bd945a6c planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
315
f181bd945a6c planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
316 ## Get the differentes codons partitions
f181bd945a6c planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
317 seq_pos1, seq_pos2, seq_pos12, seq_pos3 = get_codon_position(seq) ### DEF 12 ###
2
1f8d039bd241 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit 44a89d5eeb82789bfc643b33c11f391281b6374b
abims-sbr
parents: 1
diff changeset
318
1
f181bd945a6c planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
319 #print seq FASTA FORMAT
f181bd945a6c planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
320 OUT1.write(">%s\n" %seq_name)
f181bd945a6c planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
321 OUT1.write("%s\n" %seq)
f181bd945a6c planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
322 OUT1_pos12.write(">%s\n" %seq_name)
f181bd945a6c planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
323 OUT1_pos12.write("%s\n" %seq_pos12)
f181bd945a6c planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
324 OUT1_pos3.write(">%s\n" %seq_name)
f181bd945a6c planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
325 OUT1_pos3.write("%s\n" %seq_pos3)
2
1f8d039bd241 planemo upload for repository https://github.com/abims-sbr/adaptsearch commit 44a89d5eeb82789bfc643b33c11f391281b6374b
abims-sbr
parents: 1
diff changeset
326
1
f181bd945a6c planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
327 #print seq PHYLIP FORMAT
f181bd945a6c planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
328 OUT2.write("%s\n" %seq_name)
f181bd945a6c planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
329 OUT2.write("%s\n" %seq)
f181bd945a6c planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
330 OUT2_pos12.write("%s\n" %seq_name)
f181bd945a6c planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
331 OUT2_pos12.write("%s\n" %seq_pos12)
f181bd945a6c planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
332 OUT2_pos3.write("%s\n" %seq_name)
f181bd945a6c planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
333 OUT2_pos3.write("%s\n" %seq_pos3)
f181bd945a6c planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
334
f181bd945a6c planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
335 #print seq NEXUS FORMAT
f181bd945a6c planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
336 OUT3.write("%s" %seq_name)
f181bd945a6c planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
337 OUT3.write(" %s\n" %seq)
f181bd945a6c planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
338 OUT3_pos12.write("%s" %seq_name)
f181bd945a6c planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
339 OUT3_pos12.write(" %s\n" %seq_pos12)
f181bd945a6c planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
340 OUT3_pos3.write("%s" %seq_name)
f181bd945a6c planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
341 OUT3_pos3.write(" %s\n" %seq_pos3)
f181bd945a6c planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
342
f181bd945a6c planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
343
f181bd945a6c planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
344 OUT3.write("\t;\n")
f181bd945a6c planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
345 OUT3.write("End;\n")
f181bd945a6c planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
346 OUT3_pos12.write("\t;\n")
f181bd945a6c planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
347 OUT3_pos12.write("End;\n")
f181bd945a6c planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
348 OUT3_pos3.write("\t;\n")
f181bd945a6c planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
349 OUT3_pos3.write("End;\n")
f181bd945a6c planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
350
f181bd945a6c planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
351 OUT1.close()
f181bd945a6c planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
352 OUT2.close()
f181bd945a6c planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
353 OUT3.close()
f181bd945a6c planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
354 OUT1_pos12.close()
f181bd945a6c planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
355 OUT2_pos12.close()
f181bd945a6c planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
356 OUT3_pos12.close()
f181bd945a6c planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
357 OUT1_pos3.close()
f181bd945a6c planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
358 OUT2_pos3.close()
f181bd945a6c planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
359 OUT3_pos3.close()
f181bd945a6c planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
360
f181bd945a6c planemo upload for repository https://github.com/abims-sbr/adaptsearch commit ab76075e541dd7ece1090f6b55ca508ec0fde39d
lecorguille
parents:
diff changeset
361 print "\n\n\n******************** RAxML RUN ********************\n"