annotate otu.py @ 4:bb29ae8708b5 draft default tip

planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 7036ce0e06b6dc64332b1a5642fc58928523c5c6
author iuc
date Tue, 13 May 2025 11:52:17 +0000
parents f8ebd1e802d7
children
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
rev   line source
0
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
1 #!/usr/bin/env python3
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
2
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
3
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
4 # Name: virAnnot_otu
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
5 # Author: Marie Lefebvre - INRAE
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
6 # Reuirements: Ete3 toolkit and external apps
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
7
4
bb29ae8708b5 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 7036ce0e06b6dc64332b1a5642fc58928523c5c6
iuc
parents: 3
diff changeset
8 """Create viral OTUs based on RPS and Blast annotations"""
0
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
9
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
10 import argparse
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
11 import csv
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
12 import logging as log
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
13 import os
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
14 import random
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
15 import re
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
16
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
17 import pandas as pd
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
18 import xlsxwriter
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
19 from Bio import SeqIO
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
20 from Bio.Align.Applications import ClustalOmegaCommandline
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
21 from ete3 import NodeStyle, SeqGroup, SeqMotifFace, Tree, TreeStyle
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
22
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
23
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
24 def main():
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
25 """
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
26 1 - retrieve info (sequence, query_id, taxo) from RPS file
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
27 2 - align protein sequences of the same domain, calculate
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
28 matrix of distances, generate trees
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
29 3 - get statistics (read number) per otu
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
30 4 - create HTML report
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
31 """
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
32 options = _set_options()
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
33 _set_log_level(options.verbosity)
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
34 hits_collection = _cut_sequence(options)
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
35 _align_sequences(options, hits_collection)
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
36 _get_stats(options, hits_collection)
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
37 _create_html(options, hits_collection)
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
38
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
39
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
40 def _cut_sequence(options):
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
41 """
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
42 Retrieve viral hits and sequences from RPS files
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
43 """
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
44 log.info("Cut sequences")
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
45 i = 0 # keep track of iterations over rps files to use the corresponding fasta file
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
46 collection = {}
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
47 options.rps.sort()
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
48 for rps_file in options.rps:
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
49 log.debug("Reading rps file " + str(rps_file))
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
50 with open(rps_file[0], 'r') as rps_current_file:
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
51 rps_reader = csv.reader(rps_current_file, delimiter='\t')
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
52 headers = 0
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
53 for row in rps_reader:
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
54 if headers == 0:
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
55 # headers
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
56 headers += 1
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
57 else:
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
58 if row[1] == "no_hit":
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
59 pass
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
60 else:
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
61 query_id = row[0]
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
62 cdd_id = row[2]
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
63 startQ = int(row[5])
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
64 endQ = int(row[6])
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
65 frame = float(row[7])
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
66 description = row[8]
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
67 superkingdom = row[9]
4
bb29ae8708b5 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 7036ce0e06b6dc64332b1a5642fc58928523c5c6
iuc
parents: 3
diff changeset
68 try:
bb29ae8708b5 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 7036ce0e06b6dc64332b1a5642fc58928523c5c6
iuc
parents: 3
diff changeset
69 pident = row[10]
bb29ae8708b5 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 7036ce0e06b6dc64332b1a5642fc58928523c5c6
iuc
parents: 3
diff changeset
70 except IndexError:
bb29ae8708b5 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 7036ce0e06b6dc64332b1a5642fc58928523c5c6
iuc
parents: 3
diff changeset
71 log.info(rps_file[0])
bb29ae8708b5 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 7036ce0e06b6dc64332b1a5642fc58928523c5c6
iuc
parents: 3
diff changeset
72 log.info(row)
0
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
73 match = re.search("Viruses", superkingdom)
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
74 # if contig is viral then retrieve sequence
4
bb29ae8708b5 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 7036ce0e06b6dc64332b1a5642fc58928523c5c6
iuc
parents: 3
diff changeset
75 if match and float(pident) >= options.viral_portion:
0
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
76 options.fasta.sort()
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
77 seq = _retrieve_fasta_seq(options.fasta[i][0], query_id)
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
78 seq_length = len(seq)
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
79 if endQ < seq_length:
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
80 seq = seq[startQ - 1:endQ]
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
81 else:
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
82 seq = seq[startQ - 1:seq_length]
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
83 if frame < 0:
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
84 seq = seq.reverse_complement()
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
85 prot = seq.translate()
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
86 if len(prot) >= options.min_protein_length:
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
87 log.debug("Add " + query_id + " to collection")
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
88 if cdd_id not in collection:
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
89 collection[cdd_id] = {}
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
90 collection[cdd_id][query_id] = {}
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
91 collection[cdd_id][query_id]["nuccleotide"] = seq
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
92 collection[cdd_id][query_id]["protein"] = prot
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
93 collection[cdd_id][query_id]["full_description"] = description
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
94 if options.blast is not None:
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
95 options.blast.sort()
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
96 with open(options.blast[i][0], 'r') as blast_current_file:
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
97 blast_reader = csv.reader(blast_current_file, delimiter='\t')
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
98 for b_query in blast_reader:
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
99 if b_query[1] == query_id:
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
100 collection[cdd_id][query_id]["nb"] = b_query[2]
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
101 if len(b_query) > 10:
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
102 collection[cdd_id][query_id]["taxonomy"] = b_query[14]
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
103 else:
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
104 collection[cdd_id][query_id]["taxonomy"] = "Unknown"
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
105 else:
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
106 if "nb" not in collection[cdd_id][query_id]:
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
107 collection[cdd_id][query_id]["nb"] = 0
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
108 if "taxonomy" not in collection[cdd_id][query_id]:
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
109 collection[cdd_id][query_id]["taxonomy"] = "Unknown"
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
110 else:
4
bb29ae8708b5 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 7036ce0e06b6dc64332b1a5642fc58928523c5c6
iuc
parents: 3
diff changeset
111 log.debug("No blast file")
0
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
112 collection[cdd_id][query_id]["taxonomy"] = "Unknown"
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
113 collection[cdd_id][query_id]["nb"] = 0
4
bb29ae8708b5 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 7036ce0e06b6dc64332b1a5642fc58928523c5c6
iuc
parents: 3
diff changeset
114 # keep pfamXXX and RdRp 1
bb29ae8708b5 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 7036ce0e06b6dc64332b1a5642fc58928523c5c6
iuc
parents: 3
diff changeset
115 collection[cdd_id]["short_description"] = description.split(",")[0] + description.split(",")[1]
0
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
116 collection[cdd_id]["full_description"] = description
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
117 i += 1
4
bb29ae8708b5 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 7036ce0e06b6dc64332b1a5642fc58928523c5c6
iuc
parents: 3
diff changeset
118 if options.merge_rdrp == "yes":
bb29ae8708b5 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 7036ce0e06b6dc64332b1a5642fc58928523c5c6
iuc
parents: 3
diff changeset
119 rdrp_list = ["pfam00680", "pfam02123", "pfam00978", "pfam00998"]
bb29ae8708b5 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 7036ce0e06b6dc64332b1a5642fc58928523c5c6
iuc
parents: 3
diff changeset
120 collection["RdRp_merge"] = {}
bb29ae8708b5 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 7036ce0e06b6dc64332b1a5642fc58928523c5c6
iuc
parents: 3
diff changeset
121 for cdd_id in collection:
bb29ae8708b5 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 7036ce0e06b6dc64332b1a5642fc58928523c5c6
iuc
parents: 3
diff changeset
122 if cdd_id in rdrp_list and cdd_id != "RdRp_merge":
bb29ae8708b5 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 7036ce0e06b6dc64332b1a5642fc58928523c5c6
iuc
parents: 3
diff changeset
123 log.info("Add " + cdd_id + " in merge")
bb29ae8708b5 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 7036ce0e06b6dc64332b1a5642fc58928523c5c6
iuc
parents: 3
diff changeset
124 for query_id in collection[cdd_id]:
bb29ae8708b5 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 7036ce0e06b6dc64332b1a5642fc58928523c5c6
iuc
parents: 3
diff changeset
125 if query_id not in collection["RdRp_merge"]:
bb29ae8708b5 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 7036ce0e06b6dc64332b1a5642fc58928523c5c6
iuc
parents: 3
diff changeset
126 collection["RdRp_merge"][query_id] = {}
bb29ae8708b5 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 7036ce0e06b6dc64332b1a5642fc58928523c5c6
iuc
parents: 3
diff changeset
127 collection["RdRp_merge"][query_id] = collection[cdd_id][query_id]
0
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
128 return collection
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
129
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
130
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
131 def _retrieve_fasta_seq(fasta_file, query_id):
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
132 """
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
133 From fasta file retrieve specific sequence with id
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
134 """
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
135 contigs_list = SeqIO.to_dict(SeqIO.parse(open(fasta_file), 'fasta'))
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
136 try:
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
137 seq = contigs_list[query_id].seq
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
138 except KeyError:
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
139 print("KeyError for " + query_id + " file " + fasta_file)
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
140 else:
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
141 return seq
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
142
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
143
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
144 def _create_tree(tree, fasta, out, color):
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
145 """
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
146 Create phylogenic tree from multiple alignments
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
147 """
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
148 try:
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
149 f = open(tree, 'r')
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
150 except IOError:
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
151 log.info("Unknown file: " + tree + ". You may have less than 2 sequences to align.")
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
152 return
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
153
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
154 line = ""
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
155 for word in f:
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
156 line += word.strip()
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
157
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
158 f.close()
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
159 seqs = SeqGroup(fasta, format="fasta")
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
160 t = Tree(tree)
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
161 ts = TreeStyle()
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
162 ts.show_branch_length = True
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
163 colors = _parse_color_file(color)
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
164 node_names = t.get_leaf_names()
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
165 for name in node_names:
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
166 seq = seqs.get_seq(name)
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
167 seqFace = SeqMotifFace(seq, seq_format="()")
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
168 node = t.get_leaves_by_name(name)
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
169 for i in range(0, len(node)):
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
170 if name in colors:
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
171 ns = NodeStyle()
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
172 ns['bgcolor'] = colors[name]
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
173 node[i].set_style(ns)
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
174 node[i].add_face(seqFace, 0, 'aligned')
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
175
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
176 t.render(out, tree_style=ts)
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
177
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
178
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
179 def _parse_color_file(file):
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
180 fh = open(file)
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
181 reader = csv.reader(fh, delimiter="\t")
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
182 data = list(reader)
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
183 colors = {}
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
184 for i in range(0, len(data)):
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
185 colors[data[i][0]] = data[i][1]
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
186
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
187 return colors
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
188
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
189
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
190 def _align_sequences(options, hits_collection):
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
191 """
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
192 Align hit sequences with pfam reference
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
193 """
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
194 log.info("Align sequences")
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
195 if not os.path.exists(options.output):
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
196 os.mkdir(options.output)
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
197 color_by_sample = {}
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
198 for cdd_id in hits_collection:
4
bb29ae8708b5 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 7036ce0e06b6dc64332b1a5642fc58928523c5c6
iuc
parents: 3
diff changeset
199 log.info("align seq for " + cdd_id)
bb29ae8708b5 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 7036ce0e06b6dc64332b1a5642fc58928523c5c6
iuc
parents: 3
diff changeset
200 if cdd_id == "RdRp_merge":
bb29ae8708b5 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 7036ce0e06b6dc64332b1a5642fc58928523c5c6
iuc
parents: 3
diff changeset
201 cdd_output = options.output + "/" + cdd_id
bb29ae8708b5 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 7036ce0e06b6dc64332b1a5642fc58928523c5c6
iuc
parents: 3
diff changeset
202 else:
bb29ae8708b5 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 7036ce0e06b6dc64332b1a5642fc58928523c5c6
iuc
parents: 3
diff changeset
203 cdd_output = options.output + "/" + hits_collection[cdd_id]["short_description"].replace(" ", "_")
0
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
204 if not os.path.exists(cdd_output):
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
205 os.mkdir(cdd_output)
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
206 if os.path.exists(cdd_output + "/seq_to_align.fasta"):
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
207 os.remove(cdd_output + "/seq_to_align.fasta")
3
f8ebd1e802d7 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 16701bfbffd605805e847897799251ab748f559f
iuc
parents: 0
diff changeset
208 if os.path.exists(cdd_output + "/seq_nucc.fasta"):
f8ebd1e802d7 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 16701bfbffd605805e847897799251ab748f559f
iuc
parents: 0
diff changeset
209 os.remove(cdd_output + "/seq_nucc.fasta")
0
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
210 file_seq_to_align = cdd_output + "/seq_to_align.fasta"
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
211 file_color_config = cdd_output + "/color_config.txt"
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
212 f = open(file_seq_to_align, "a")
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
213 f_c = open(file_color_config, "w+")
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
214 log.info("Writing to " + file_seq_to_align)
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
215 count = 0 # count number of contig per domain
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
216 for query_id in hits_collection[cdd_id]:
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
217 if query_id not in ["short_description", "full_description"]:
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
218 sample = query_id.split("_")[0] # get sample from SAMPLE_IdCONTIG
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
219 sample_color = "#" + ''.join([random.choice('ABCDEF0123456789') for i in range(6)])
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
220 # same color for each contig of the same sample
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
221 if sample not in color_by_sample.keys():
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
222 color_by_sample[sample] = sample_color
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
223 f.write(">" + query_id + "\n")
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
224 f.write(str(hits_collection[cdd_id][query_id]["protein"]) + "\n")
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
225 f_c.write(query_id + '\t' + color_by_sample[sample] + '\n')
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
226 count += 1
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
227 f.close()
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
228 f_c.close()
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
229 file_seq_aligned = cdd_output + '/seq_aligned.final_tree.fa'
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
230 tree_file = cdd_output + '/tree.dnd'
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
231 file_cluster = cdd_output + '/otu_cluster.csv'
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
232 # create alignment for domain with more than 1 contigs
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
233 if count > 1:
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
234 log.info("Run clustal omega...")
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
235 clustalo_cmd = ClustalOmegaCommandline("clustalo", infile=file_seq_to_align, outfile=file_seq_aligned,
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
236 guidetree_out=tree_file, seqtype="protein", force=True)
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
237 log.debug(clustalo_cmd)
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
238 stdout, stderr = clustalo_cmd()
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
239 log.debug(stdout + stderr)
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
240
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
241 # create tree plot with colors
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
242 file_matrix = cdd_output + "/identity_matrix.csv"
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
243 log.info("Create tree...")
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
244 _create_tree(tree_file, file_seq_aligned, tree_file + '.png', file_color_config)
4
bb29ae8708b5 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 7036ce0e06b6dc64332b1a5642fc58928523c5c6
iuc
parents: 3
diff changeset
245 _compute_pairwise_distance(file_seq_aligned, file_matrix, cdd_id)
0
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
246 log.info("Retrieve OTUs...")
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
247 # if os.path.exists(file_cluster):
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
248 # os.remove(file_cluster)
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
249 otu_cmd = os.path.join(options.tool_path, 'seek_otu.R') + ' ' + file_matrix + ' ' + file_cluster + ' ' + str(options.perc)
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
250 log.debug(otu_cmd)
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
251 os.system(otu_cmd)
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
252 # only one contig
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
253 else:
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
254 mv_cmd = 'cp ' + file_seq_to_align + ' ' + file_seq_aligned
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
255 log.debug(mv_cmd)
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
256 os.system(mv_cmd)
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
257
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
258 f = open(file_cluster, "w+")
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
259 f.write('OTU_1,1,' + list(hits_collection[cdd_id].keys())[0] + ',')
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
260 f.close()
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
261
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
262
4
bb29ae8708b5 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 7036ce0e06b6dc64332b1a5642fc58928523c5c6
iuc
parents: 3
diff changeset
263 def _compute_pairwise_distance(file_seq_aligned, file_matrix, cdd_id):
0
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
264 """
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
265 Calculate paiwise distance between aligned protein sequences
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
266 from a cdd_id
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
267 """
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
268 log.info("Compute pairwise distance of " + cdd_id)
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
269 matrix = {}
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
270 for k1 in SeqIO.parse(file_seq_aligned, "fasta"):
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
271 row = []
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
272 for k2 in SeqIO.parse(file_seq_aligned, "fasta"):
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
273 identic = 0
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
274 compared = 0
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
275 keep_pos = 0
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
276 for base in k1:
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
277 base2 = k2[keep_pos]
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
278 # mutation, next
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
279 if base == 'X' or base2 == 'X':
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
280 keep_pos += 1
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
281 continue
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
282 # gap in both sequences, next
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
283 if base == '-' and base2 == '-':
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
284 keep_pos += 1
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
285 continue
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
286 # gap in one of the sequence, next
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
287 if base == '-' or base2 == '-':
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
288 keep_pos += 1
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
289 continue
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
290 # identity
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
291 if base == base2:
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
292 identic += 1
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
293 compared += 1
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
294 keep_pos += 1
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
295 # set minimum overlap to 20
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
296 if compared == 0 or compared < 20:
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
297 percentIdentity = 0
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
298 else:
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
299 percentIdentity = (identic / compared) * 100
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
300 row.append(percentIdentity)
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
301 matrix[k1.id] = row
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
302 log.debug("Write " + file_matrix)
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
303 f = open(file_matrix, "w+")
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
304 for row in matrix:
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
305 f.write(row + ',' + ', '.join(map(str, matrix[row])) + "\n")
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
306 f.close()
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
307
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
308
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
309 def _get_stats(options, hits_collection):
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
310 """
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
311 Retrieve annotation and number of read
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
312 for each OTUs
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
313 """
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
314 file_xlsx = options.output + '/otu_stats.xlsx' # Create a workbook
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
315 workbook = xlsxwriter.Workbook(file_xlsx)
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
316 log.info("Writing stats to " + file_xlsx)
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
317 for cdd_id in hits_collection:
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
318 otu_collection = {}
4
bb29ae8708b5 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 7036ce0e06b6dc64332b1a5642fc58928523c5c6
iuc
parents: 3
diff changeset
319 if cdd_id == "RdRp_merge":
bb29ae8708b5 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 7036ce0e06b6dc64332b1a5642fc58928523c5c6
iuc
parents: 3
diff changeset
320 cdd_output = options.output + "/" + cdd_id
bb29ae8708b5 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 7036ce0e06b6dc64332b1a5642fc58928523c5c6
iuc
parents: 3
diff changeset
321 worksheet = workbook.add_worksheet(cdd_id)
bb29ae8708b5 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 7036ce0e06b6dc64332b1a5642fc58928523c5c6
iuc
parents: 3
diff changeset
322 else:
bb29ae8708b5 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 7036ce0e06b6dc64332b1a5642fc58928523c5c6
iuc
parents: 3
diff changeset
323
bb29ae8708b5 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 7036ce0e06b6dc64332b1a5642fc58928523c5c6
iuc
parents: 3
diff changeset
324 cdd_output = options.output + "/" + hits_collection[cdd_id]["short_description"].replace(" ", "_")
bb29ae8708b5 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 7036ce0e06b6dc64332b1a5642fc58928523c5c6
iuc
parents: 3
diff changeset
325 worksheet = workbook.add_worksheet(hits_collection[cdd_id]["short_description"]) # add a worksheet
0
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
326 file_cluster = cdd_output + '/otu_cluster.csv'
3
f8ebd1e802d7 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 16701bfbffd605805e847897799251ab748f559f
iuc
parents: 0
diff changeset
327 file_fasta_nucc = cdd_output + '/representative_nucc.fasta'
0
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
328 with open(file_cluster, 'r') as clust:
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
329 otu_reader = csv.reader(clust, delimiter=',')
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
330 samples_list = []
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
331 for row in otu_reader:
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
332 contigs_list = row[2:len(row) - 1] # remove last empty column
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
333 otu_collection[row[0]] = {} # key -> otu number
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
334 otu_collection[row[0]]['contigs_list'] = contigs_list
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
335 for contig in contigs_list:
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
336 sample = contig.split('_')[0]
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
337 samples_list.append(sample) if sample not in samples_list else samples_list
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
338 if sample not in otu_collection[row[0]]:
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
339 otu_collection[row[0]][sample] = {}
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
340 otu_collection[row[0]][sample][contig] = {}
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
341 # add read number of the contig and annotation
4
bb29ae8708b5 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 7036ce0e06b6dc64332b1a5642fc58928523c5c6
iuc
parents: 3
diff changeset
342 if contig in hits_collection[cdd_id]:
bb29ae8708b5 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 7036ce0e06b6dc64332b1a5642fc58928523c5c6
iuc
parents: 3
diff changeset
343 if 'nb' in hits_collection[cdd_id][contig]:
bb29ae8708b5 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 7036ce0e06b6dc64332b1a5642fc58928523c5c6
iuc
parents: 3
diff changeset
344 otu_collection[row[0]][sample][contig]['nb'] = hits_collection[cdd_id][contig]["nb"]
bb29ae8708b5 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 7036ce0e06b6dc64332b1a5642fc58928523c5c6
iuc
parents: 3
diff changeset
345 else:
bb29ae8708b5 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 7036ce0e06b6dc64332b1a5642fc58928523c5c6
iuc
parents: 3
diff changeset
346 otu_collection[row[0]][sample][contig]['nb'] = 0
bb29ae8708b5 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 7036ce0e06b6dc64332b1a5642fc58928523c5c6
iuc
parents: 3
diff changeset
347 if 'taxonomy' in hits_collection[cdd_id][contig]:
bb29ae8708b5 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 7036ce0e06b6dc64332b1a5642fc58928523c5c6
iuc
parents: 3
diff changeset
348 otu_collection[row[0]][sample][contig]['taxonomy'] = hits_collection[cdd_id][contig]["taxonomy"]
bb29ae8708b5 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 7036ce0e06b6dc64332b1a5642fc58928523c5c6
iuc
parents: 3
diff changeset
349 else:
bb29ae8708b5 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 7036ce0e06b6dc64332b1a5642fc58928523c5c6
iuc
parents: 3
diff changeset
350 otu_collection[row[0]][sample][contig]['taxonomy'] = 'unknown'
0
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
351 else:
4
bb29ae8708b5 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 7036ce0e06b6dc64332b1a5642fc58928523c5c6
iuc
parents: 3
diff changeset
352 otu_collection[row[0]][sample][contig] = {'nb': 0, 'taxonomy': 'unknown'}
0
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
353 else:
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
354 otu_collection[row[0]][sample][contig] = {}
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
355 # add read number of the contig and annotation
4
bb29ae8708b5 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 7036ce0e06b6dc64332b1a5642fc58928523c5c6
iuc
parents: 3
diff changeset
356 if contig in hits_collection[cdd_id]:
bb29ae8708b5 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 7036ce0e06b6dc64332b1a5642fc58928523c5c6
iuc
parents: 3
diff changeset
357 if 'nb' in hits_collection[cdd_id][contig]:
bb29ae8708b5 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 7036ce0e06b6dc64332b1a5642fc58928523c5c6
iuc
parents: 3
diff changeset
358 otu_collection[row[0]][sample][contig]['nb'] = hits_collection[cdd_id][contig]["nb"]
bb29ae8708b5 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 7036ce0e06b6dc64332b1a5642fc58928523c5c6
iuc
parents: 3
diff changeset
359 else:
bb29ae8708b5 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 7036ce0e06b6dc64332b1a5642fc58928523c5c6
iuc
parents: 3
diff changeset
360 otu_collection[row[0]][sample][contig]['nb'] = 0
bb29ae8708b5 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 7036ce0e06b6dc64332b1a5642fc58928523c5c6
iuc
parents: 3
diff changeset
361 if 'taxonomy' in hits_collection[cdd_id][contig]:
bb29ae8708b5 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 7036ce0e06b6dc64332b1a5642fc58928523c5c6
iuc
parents: 3
diff changeset
362 otu_collection[row[0]][sample][contig]['taxonomy'] = hits_collection[cdd_id][contig]["taxonomy"]
bb29ae8708b5 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 7036ce0e06b6dc64332b1a5642fc58928523c5c6
iuc
parents: 3
diff changeset
363 else:
bb29ae8708b5 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 7036ce0e06b6dc64332b1a5642fc58928523c5c6
iuc
parents: 3
diff changeset
364 otu_collection[row[0]][sample][contig]['taxonomy'] = 'unknown'
0
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
365 else:
4
bb29ae8708b5 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 7036ce0e06b6dc64332b1a5642fc58928523c5c6
iuc
parents: 3
diff changeset
366 otu_collection[row[0]][sample][contig] = {'nb': 0, 'taxonomy': 'unknown'}
0
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
367 if 'taxonomy' in hits_collection[cdd_id][contig]:
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
368 otu_collection[row[0]]['global_taxonomy'] = hits_collection[cdd_id][contig]["taxonomy"]
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
369 else:
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
370 otu_collection[row[0]]['global_taxonomy'] = 'unknown'
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
371
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
372 # calculate total number of reads for each sample of each OTU
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
373 for otu in otu_collection:
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
374 for sample in otu_collection[otu]:
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
375 if sample not in ['contigs_list', 'global_taxonomy']:
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
376 total_nb_read = 0
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
377 for contig in otu_collection[otu][sample]:
3
f8ebd1e802d7 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 16701bfbffd605805e847897799251ab748f559f
iuc
parents: 0
diff changeset
378 if otu_collection[otu][sample][contig]['nb'] == '':
f8ebd1e802d7 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 16701bfbffd605805e847897799251ab748f559f
iuc
parents: 0
diff changeset
379 otu_collection[otu][sample][contig]['nb'] = 0
0
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
380 total_nb_read += int(otu_collection[otu][sample][contig]['nb'])
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
381 otu_collection[otu][sample]['total_nb_read'] = total_nb_read
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
382 row = 0
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
383 column = 0
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
384 item = '#OTU_name'
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
385 worksheet.write(row, column, item)
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
386 for samp in samples_list:
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
387 column += 1
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
388 worksheet.write(row, column, samp)
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
389 worksheet.write(row, column + 1, 'taxonomy')
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
390 worksheet.write(row, column + 2, 'contigs_list')
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
391 row = 1
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
392 # column = 0
3
f8ebd1e802d7 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 16701bfbffd605805e847897799251ab748f559f
iuc
parents: 0
diff changeset
393 with open(file_fasta_nucc, "w+") as f_nucc:
f8ebd1e802d7 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 16701bfbffd605805e847897799251ab748f559f
iuc
parents: 0
diff changeset
394 for otu in otu_collection:
f8ebd1e802d7 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 16701bfbffd605805e847897799251ab748f559f
iuc
parents: 0
diff changeset
395 if isinstance(otu_collection[otu], dict):
f8ebd1e802d7 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 16701bfbffd605805e847897799251ab748f559f
iuc
parents: 0
diff changeset
396 column = 0
f8ebd1e802d7 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 16701bfbffd605805e847897799251ab748f559f
iuc
parents: 0
diff changeset
397 worksheet.write(row, column, otu)
f8ebd1e802d7 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 16701bfbffd605805e847897799251ab748f559f
iuc
parents: 0
diff changeset
398 # prepare table with 0 in each cells
f8ebd1e802d7 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 16701bfbffd605805e847897799251ab748f559f
iuc
parents: 0
diff changeset
399 for sample in otu_collection[otu]:
f8ebd1e802d7 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 16701bfbffd605805e847897799251ab748f559f
iuc
parents: 0
diff changeset
400 column = 1
f8ebd1e802d7 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 16701bfbffd605805e847897799251ab748f559f
iuc
parents: 0
diff changeset
401 for samp in samples_list:
f8ebd1e802d7 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 16701bfbffd605805e847897799251ab748f559f
iuc
parents: 0
diff changeset
402 worksheet.write(row, column, 0)
f8ebd1e802d7 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 16701bfbffd605805e847897799251ab748f559f
iuc
parents: 0
diff changeset
403 column += 1
f8ebd1e802d7 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 16701bfbffd605805e847897799251ab748f559f
iuc
parents: 0
diff changeset
404 # fill in table with nb of read for each sample and each OTU
f8ebd1e802d7 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 16701bfbffd605805e847897799251ab748f559f
iuc
parents: 0
diff changeset
405 for sample in otu_collection[otu]:
f8ebd1e802d7 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 16701bfbffd605805e847897799251ab748f559f
iuc
parents: 0
diff changeset
406 column = 1
f8ebd1e802d7 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 16701bfbffd605805e847897799251ab748f559f
iuc
parents: 0
diff changeset
407 for samp in samples_list:
f8ebd1e802d7 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 16701bfbffd605805e847897799251ab748f559f
iuc
parents: 0
diff changeset
408 if samp == sample:
f8ebd1e802d7 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 16701bfbffd605805e847897799251ab748f559f
iuc
parents: 0
diff changeset
409 worksheet.write(row, column, otu_collection[otu][sample]['total_nb_read'])
f8ebd1e802d7 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 16701bfbffd605805e847897799251ab748f559f
iuc
parents: 0
diff changeset
410 column += 1
f8ebd1e802d7 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 16701bfbffd605805e847897799251ab748f559f
iuc
parents: 0
diff changeset
411 worksheet.write(row, len(samples_list) + 1, otu_collection[otu]['global_taxonomy'].replace(';', ' '))
f8ebd1e802d7 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 16701bfbffd605805e847897799251ab748f559f
iuc
parents: 0
diff changeset
412 worksheet.write(row, len(samples_list) + 2, ",".join(otu_collection[otu]['contigs_list']))
f8ebd1e802d7 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 16701bfbffd605805e847897799251ab748f559f
iuc
parents: 0
diff changeset
413 row += 1
f8ebd1e802d7 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 16701bfbffd605805e847897799251ab748f559f
iuc
parents: 0
diff changeset
414 f_nucc.write(">" + cdd_id + "_" + otu + "_" + otu_collection[otu]['contigs_list'][0] + "\n")
f8ebd1e802d7 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 16701bfbffd605805e847897799251ab748f559f
iuc
parents: 0
diff changeset
415 f_nucc.write(str(hits_collection[cdd_id][otu_collection[otu]['contigs_list'][0]]['nuccleotide']) + "\n")
0
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
416 workbook.close()
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
417 read_file = pd.ExcelFile(file_xlsx)
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
418 for sheet in read_file.sheet_names:
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
419 cluster_nb_reads_file = options.output + "/" + sheet.replace(" ", "_") + "/cluster_nb_reads_files.tab"
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
420 data_xls = pd.read_excel(file_xlsx, sheet, dtype=str, index_col=None)
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
421 data_xls.to_csv(cluster_nb_reads_file, encoding='utf-8', index=False, sep='\t')
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
422
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
423
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
424 def _create_html(options, hits_collection):
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
425 """
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
426 Create HTML file with all results
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
427 """
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
428 # create mapping file with all informations to use to create HTML report
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
429 map_file_path = options.output + "/map.txt"
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
430 if os.path.exists(map_file_path):
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
431 os.remove(map_file_path)
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
432
3
f8ebd1e802d7 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 16701bfbffd605805e847897799251ab748f559f
iuc
parents: 0
diff changeset
433 with open(map_file_path, "w+") as map_file:
f8ebd1e802d7 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 16701bfbffd605805e847897799251ab748f559f
iuc
parents: 0
diff changeset
434 headers = ['#cdd_id', 'align_files', 'tree_files', 'cluster_files', 'cluster_nb_reads_files', 'pairwise_files', 'description', 'full_description\n']
f8ebd1e802d7 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 16701bfbffd605805e847897799251ab748f559f
iuc
parents: 0
diff changeset
435 map_file.write("\t".join(headers))
f8ebd1e802d7 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 16701bfbffd605805e847897799251ab748f559f
iuc
parents: 0
diff changeset
436 for cdd_id in hits_collection:
4
bb29ae8708b5 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 7036ce0e06b6dc64332b1a5642fc58928523c5c6
iuc
parents: 3
diff changeset
437 if cdd_id == "RdRp_merge":
bb29ae8708b5 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 7036ce0e06b6dc64332b1a5642fc58928523c5c6
iuc
parents: 3
diff changeset
438 cdd_output = "RdRp_merge"
bb29ae8708b5 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 7036ce0e06b6dc64332b1a5642fc58928523c5c6
iuc
parents: 3
diff changeset
439 else:
bb29ae8708b5 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 7036ce0e06b6dc64332b1a5642fc58928523c5c6
iuc
parents: 3
diff changeset
440 cdd_output = hits_collection[cdd_id]["short_description"].replace(" ", "_")
3
f8ebd1e802d7 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 16701bfbffd605805e847897799251ab748f559f
iuc
parents: 0
diff changeset
441 short_description = cdd_output
f8ebd1e802d7 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 16701bfbffd605805e847897799251ab748f559f
iuc
parents: 0
diff changeset
442 file_seq_aligned = cdd_output + '/seq_aligned.final_tree.fa'
f8ebd1e802d7 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 16701bfbffd605805e847897799251ab748f559f
iuc
parents: 0
diff changeset
443 tree_file = cdd_output + '/tree.dnd.png'
f8ebd1e802d7 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 16701bfbffd605805e847897799251ab748f559f
iuc
parents: 0
diff changeset
444 file_cluster = cdd_output + '/otu_cluster.csv'
f8ebd1e802d7 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 16701bfbffd605805e847897799251ab748f559f
iuc
parents: 0
diff changeset
445 file_matrix = cdd_output + "/identity_matrix.csv"
f8ebd1e802d7 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 16701bfbffd605805e847897799251ab748f559f
iuc
parents: 0
diff changeset
446 cluster_nb_reads_files = cdd_output + "/cluster_nb_reads_files.tab"
f8ebd1e802d7 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 16701bfbffd605805e847897799251ab748f559f
iuc
parents: 0
diff changeset
447 map_file.write(cdd_id + "\t" + file_seq_aligned + "\t" + tree_file + "\t")
f8ebd1e802d7 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 16701bfbffd605805e847897799251ab748f559f
iuc
parents: 0
diff changeset
448 map_file.write(file_cluster + "\t" + cluster_nb_reads_files + "\t" + file_matrix + "\t")
f8ebd1e802d7 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 16701bfbffd605805e847897799251ab748f559f
iuc
parents: 0
diff changeset
449 map_file.write(short_description + "\t" + hits_collection[cdd_id]["full_description"] + "\n")
0
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
450 log.info("Writing HTML report")
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
451 html_cmd = os.path.join(options.tool_path, 'rps2tree_html.py') + ' -m ' + map_file_path + ' -o ' + options.output
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
452 log.debug(html_cmd)
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
453 os.system(html_cmd)
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
454
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
455
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
456 def _set_options():
4
bb29ae8708b5 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 7036ce0e06b6dc64332b1a5642fc58928523c5c6
iuc
parents: 3
diff changeset
457 """
bb29ae8708b5 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 7036ce0e06b6dc64332b1a5642fc58928523c5c6
iuc
parents: 3
diff changeset
458 Set parameters
bb29ae8708b5 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 7036ce0e06b6dc64332b1a5642fc58928523c5c6
iuc
parents: 3
diff changeset
459 """
0
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
460 parser = argparse.ArgumentParser()
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
461 parser.add_argument('-b', '--blast', help='TAB blast file from blast2ecsv module.', action='append', required=False, dest='blast', nargs='+')
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
462 parser.add_argument('-r', '--rps', help='TAB rpsblast file from rps2ecsv module.', action='append', required=True, dest='rps', nargs='+')
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
463 parser.add_argument('-f', '--fasta', help='FASTA file with contigs', action='append', required=True, dest='fasta', nargs='+')
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
464 parser.add_argument('-p', '--percentage', help='Percentage similarity threshold for OTUs cutoff.', action='store', type=int, default=90, dest='perc')
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
465 parser.add_argument('-vp', '--viral_portion', help='Minimun portion of viral sequences in RPS domain to be included.', action='store', type=float, default=0.3, dest='viral_portion')
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
466 parser.add_argument('-mpl', '--min_protein_length', help='Minimum query protein length.', action='store', type=int, default=100, dest='min_protein_length')
4
bb29ae8708b5 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 7036ce0e06b6dc64332b1a5642fc58928523c5c6
iuc
parents: 3
diff changeset
467 parser.add_argument('-m', '--merge_rdrp', help='Merge RdRp1, 2, 3 and 4 to create otu on it.', action='store', type=str, default="no", dest='merge_rdrp')
0
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
468 parser.add_argument('-tp', '--tool_path', help='Path to otu_seek.R', action='store', type=str, default='./', dest='tool_path')
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
469 parser.add_argument('-o', '--out', help='The output directory', action='store', type=str, default='./Rps2tree_OTU', dest='output')
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
470 parser.add_argument('-rgb', '--rgb-conf', help='Color palette for contigs coloration', action='store', type=str, default='rgb.txt', dest='file_rgb')
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
471 parser.add_argument('-v', '--verbosity', help='Verbose level', action='store', type=int, choices=[1, 2, 3, 4], default=1)
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
472 args = parser.parse_args()
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
473 return args
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
474
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
475
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
476 def _set_log_level(verbosity):
4
bb29ae8708b5 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 7036ce0e06b6dc64332b1a5642fc58928523c5c6
iuc
parents: 3
diff changeset
477 """
bb29ae8708b5 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 7036ce0e06b6dc64332b1a5642fc58928523c5c6
iuc
parents: 3
diff changeset
478 Debbug
bb29ae8708b5 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 7036ce0e06b6dc64332b1a5642fc58928523c5c6
iuc
parents: 3
diff changeset
479 """
0
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
480 if verbosity == 1:
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
481 log_format = '%(asctime)s %(levelname)-8s %(message)s'
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
482 log.basicConfig(level=log.INFO, format=log_format)
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
483 elif verbosity == 3:
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
484 log_format = '%(filename)s:%(lineno)s - %(asctime)s %(levelname)-8s %(message)s'
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
485 log.basicConfig(level=log.DEBUG, format=log_format)
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
486
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
487
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
488 if __name__ == "__main__":
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
489 main()