Mercurial > repos > iuc > virannot_blast2tsv

annotate blast2tsv.py @ 3:f8ebd1e802d7 draft default tip

Find changesets by keywords (author, files, the commit message), revision number or hash, or revset expression.
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 16701bfbffd605805e847897799251ab748f559f
author iuc
date Sun, 08 Sep 2024 14:09:19 +0000
parents 77c3ef9b0ed7
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: blast2tsv
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
5 # Author(s): Sebastien Theil, 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 # Aims: Convert blast xml output to tsv and add taxonomy
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
7
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
8
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
9 import argparse
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
10 import csv
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
11 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
12 import os
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
13
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
14 from Bio import Entrez
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
15 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
16 from Bio.Blast import NCBIXML
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
17 from ete3 import NCBITaxa
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
18
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
19 ncbi = NCBITaxa()
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
20
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
21
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
22 def main():
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
23 options = _set_options()
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
24 _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
25 hits = _read_xml(options)
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
26 _write_tsv(options, hits)
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
27
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
28
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
29 def _guess_database(accession):
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
30 """Guess the correct database for querying based off the format of the accession"""
2
77c3ef9b0ed7 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit ab5e1189217b6ed5f1c5d7c5ff6b79b6a4c18cff
iuc
parents: 0
diff changeset
31 if accession.isdigit():
77c3ef9b0ed7 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit ab5e1189217b6ed5f1c5d7c5ff6b79b6a4c18cff
iuc
parents: 0
diff changeset
32 db = 'taxonomy'
77c3ef9b0ed7 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit ab5e1189217b6ed5f1c5d7c5ff6b79b6a4c18cff
iuc
parents: 0
diff changeset
33 else:
77c3ef9b0ed7 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit ab5e1189217b6ed5f1c5d7c5ff6b79b6a4c18cff
iuc
parents: 0
diff changeset
34 database_mappings_refseq = {'AC': 'nuccore', 'NC': 'nuccore', 'NG': 'nuccore',
77c3ef9b0ed7 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit ab5e1189217b6ed5f1c5d7c5ff6b79b6a4c18cff
iuc
parents: 0
diff changeset
35 'NT': 'nuccore', 'NW': 'nuccore', 'NZ': 'nuccore',
77c3ef9b0ed7 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit ab5e1189217b6ed5f1c5d7c5ff6b79b6a4c18cff
iuc
parents: 0
diff changeset
36 'AP': 'protein', 'NP': 'protein', 'YP': 'protein',
77c3ef9b0ed7 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit ab5e1189217b6ed5f1c5d7c5ff6b79b6a4c18cff
iuc
parents: 0
diff changeset
37 'XP': 'protein', 'WP': 'protein', 'OX': 'nuccore'}
77c3ef9b0ed7 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit ab5e1189217b6ed5f1c5d7c5ff6b79b6a4c18cff
iuc
parents: 0
diff changeset
38 try:
77c3ef9b0ed7 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit ab5e1189217b6ed5f1c5d7c5ff6b79b6a4c18cff
iuc
parents: 0
diff changeset
39 db = database_mappings_refseq[accession[0:2]]
77c3ef9b0ed7 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit ab5e1189217b6ed5f1c5d7c5ff6b79b6a4c18cff
iuc
parents: 0
diff changeset
40 except KeyError:
77c3ef9b0ed7 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit ab5e1189217b6ed5f1c5d7c5ff6b79b6a4c18cff
iuc
parents: 0
diff changeset
41 db = 'nuccore'
77c3ef9b0ed7 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit ab5e1189217b6ed5f1c5d7c5ff6b79b6a4c18cff
iuc
parents: 0
diff changeset
42 log.warning("DB not found for " + accession + ". Set to nuccore.")
77c3ef9b0ed7 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit ab5e1189217b6ed5f1c5d7c5ff6b79b6a4c18cff
iuc
parents: 0
diff changeset
43 return db
0
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
44
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
45
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
46 def _read_xml(options):
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
47 """
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
48 Parse XML blast results file
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
49 Keep only the first hit
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
50 """
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
51 log.info("Read XML file.")
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
52 results = open(options.xml_file, 'r')
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
53 records = NCBIXML.parse(results)
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
54 xml_results = {}
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
55 for blast_record in records:
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
56 for aln in blast_record.alignments:
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
57 hit_count = 1
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
58 for hit in aln.hsps:
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
59 hsp = {}
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
60 if hit_count == 1:
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
61 first_hit_frame = hit.frame[1] if len(hit.frame) > 0 else 0 # strand
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
62 cumul_hit_identity = hit.identities if hit.identities else 0
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
63 cumul_hit_score = hit.bits # hit score
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
64 cumul_hit_evalue = hit.expect # evalue
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
65 cumul_hit_length = hit.align_length if hit.align_length is not None else 0
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
66 hit_count = hit_count + 1
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
67 else:
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
68 # all HSPs in different strand than 1st HSPs will be discarded.
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
69 if (first_hit_frame > 0 and hit.frame[1] > 0) or (first_hit_frame < 0 and hit.frame[1] < 0):
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
70 cumul_hit_identity = cumul_hit_identity + hit.identities
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
71 cumul_hit_length = cumul_hit_length + hit.align_length
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
72 cumul_hit_evalue = cumul_hit_evalue + hit.expect
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
73 cumul_hit_score = cumul_hit_score + hit.bits
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
74 hit_count = hit_count + 1
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
75 if hit_count == 1:
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
76 final_hit_count = hit_count
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
77 elif hit_count > 1:
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
78 final_hit_count = hit_count - 1
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
79 hsp["evalue"] = cumul_hit_evalue / final_hit_count # The smaller the E-value, the better the match
2
77c3ef9b0ed7 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit ab5e1189217b6ed5f1c5d7c5ff6b79b6a4c18cff
iuc
parents: 0
diff changeset
80 hsp["query_id"] = blast_record.query # or query_id
0
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
81 hsp["query_length"] = blast_record.query_length # length of the query
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
82 hsp["accession"] = aln.accession.replace("ref|", "")
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
83 hsp["description"] = aln.hit_def
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
84 hsp["hit_length"] = aln.length # length of the hit
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
85 hsp["hsp_length"] = hit.align_length # length of the hsp alignment
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
86 hsp["queryOverlap"] = _get_overlap_value(options.algo, hsp, 'hsp', hsp["query_length"])[0]
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
87 if cumul_hit_length == 0:
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
88 hsp["percentIdentity"] = round(cumul_hit_identity, 1) # identity percentage
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
89 else:
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
90 hsp["percentIdentity"] = round(cumul_hit_identity / cumul_hit_length * 100, 1) # identity percentage
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
91 hsp["score"] = cumul_hit_score # The higher the bit-score, the better the sequence similarity
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
92 hsp["num_hsps"] = final_hit_count
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
93 hsp["hit_cumul_length"] = cumul_hit_length
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
94 hsp["hitOverlap"] = _get_overlap_value(options.algo, hsp, 'hit', hsp["query_length"])[1]
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
95 db = _guess_database(hsp["accession"])
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
96 try:
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
97 handle = Entrez.esummary(db=db, id=hsp["accession"])
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
98 taxid = str(int(Entrez.read(handle)[0]['TaxId']))
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
99 handle.close()
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
100 log.info("Taxid found for " + hsp["accession"])
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
101 lineage = ncbi.get_lineage(taxid)
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
102 names = ncbi.get_taxid_translator(lineage)
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
103 ordered = [names[tid] for tid in lineage]
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
104 taxonomy = ordered[1:]
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
105 hsp["tax_id"] = taxid
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
106 hsp["taxonomy"] = ';'.join(taxonomy)
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
107 hsp["organism"] = taxonomy[-1]
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
108 except RuntimeError:
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
109 hsp["tax_id"] = ""
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
110 hsp["taxonomy"] = ""
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
111 hsp["organism"] = ""
2
77c3ef9b0ed7 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit ab5e1189217b6ed5f1c5d7c5ff6b79b6a4c18cff
iuc
parents: 0
diff changeset
112 log.warning(f"RuntimeError - Taxid not found for {hsp['accession']}")
77c3ef9b0ed7 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit ab5e1189217b6ed5f1c5d7c5ff6b79b6a4c18cff
iuc
parents: 0
diff changeset
113 except Exception as err:
77c3ef9b0ed7 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit ab5e1189217b6ed5f1c5d7c5ff6b79b6a4c18cff
iuc
parents: 0
diff changeset
114 hsp["tax_id"] = ""
77c3ef9b0ed7 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit ab5e1189217b6ed5f1c5d7c5ff6b79b6a4c18cff
iuc
parents: 0
diff changeset
115 hsp["taxonomy"] = ""
77c3ef9b0ed7 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit ab5e1189217b6ed5f1c5d7c5ff6b79b6a4c18cff
iuc
parents: 0
diff changeset
116 hsp["organism"] = ""
77c3ef9b0ed7 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit ab5e1189217b6ed5f1c5d7c5ff6b79b6a4c18cff
iuc
parents: 0
diff changeset
117 log.warning(f"Taxid not found for {hsp['accession']}. The error is {err}")
0
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
118 if hsp["evalue"] <= options.max_evalue and hsp["queryOverlap"] >= options.min_qov and \
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
119 hsp["hitOverlap"] >= options.min_hov and hsp["score"] >= options.min_score:
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
120 xml_results[hsp["query_id"]] = hsp
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
121 else:
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
122 xml_results[hsp["query_id"]] = [hsp["query_length"]]
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
123
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
124 return xml_results
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
125
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
126
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
127 def _get_overlap_value(algo, hsp, type, qlength):
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
128 """
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
129 Set hsp or hit overlap values for hit and query
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
130 Return array [query_overlap, hit_overlap]
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
131 """
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
132 if type == 'hsp':
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
133 q_align_len = qlength
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
134 h_align_len = hsp["hsp_length"]
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
135 else:
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
136 q_align_len = qlength
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
137 h_align_len = hsp["hit_cumul_length"]
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
138
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
139 if algo == 'BLASTX':
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
140 if q_align_len:
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
141 query_overlap = (q_align_len * 3 / q_align_len) * 100
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
142 if hsp["hit_length"]:
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
143 hit_overlap = (h_align_len / hsp["hit_length"]) * 100
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
144 elif algo == 'TBLASTN':
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
145 if q_align_len:
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
146 query_overlap = (q_align_len / q_align_len) * 100
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
147 if hsp["hit_length"]:
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
148 hit_overlap = (h_align_len * 3 / hsp["hit_length"]) * 100
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
149 elif algo == 'TBLASTX':
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
150 if q_align_len:
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
151 query_overlap = (q_align_len * 3 / hsp["hsp_length"]) * 100
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
152 if hsp["hit_length"]:
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
153 hit_overlap = (h_align_len * 3 / hsp["hit_length"]) * 100
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
154 else:
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
155 if q_align_len:
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
156 query_overlap = (q_align_len / q_align_len) * 100
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
157 if hsp["hit_length"]:
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
158 hit_overlap = (h_align_len / hsp["hit_length"]) * 100
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
159 if query_overlap is None:
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
160 query_overlap = 0
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
161 if query_overlap > 100:
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
162 query_overlap = 100
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
163 if 'hit_overlap' not in locals():
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
164 hit_overlap = 0
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
165 if hit_overlap > 100:
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
166 hit_overlap = 100
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
167
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
168 return [round(query_overlap, 0), round(hit_overlap, 0)]
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
169
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
170
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
171 def _write_tsv(options, hits):
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
172 """
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
173 Write output
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
174 """
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
175 # get a list of contig without corresponding number of mapped reads
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
176 if options.rn_file is not None:
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
177 with open(options.rn_file) as rn:
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
178 rows = (line.split('\t') for line in rn)
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
179 rn_list = {row[0]: row[1:] for row in rows}
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
180 fasta = SeqIO.to_dict(SeqIO.parse(open(options.fasta_file), 'fasta'))
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
181 headers = "#algo\tquery_id\tnb_reads\tquery_length\taccession\tdescription\torganism\tpercentIdentity\tnb_hsps\tqueryOverlap\thitOverlap\tevalue\tscore\ttax_id\ttaxonomy\tsequence\n"
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
182 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
183 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
184 tsv_file = options.output + "/blast2tsv_output.tab"
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
185 log.info("Write output file: " + tsv_file)
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
186 f = open(tsv_file, "w+")
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
187 f.write(headers)
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
188 for h in hits:
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
189 if options.rn_file is not None:
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
190 read_nb = ''.join(rn_list[h]).replace("\n", "")
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
191 else:
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
192 read_nb = ''
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
193 if len(hits[h]) > 1:
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
194 f.write(options.algo + "\t" + h + "\t" + read_nb + "\t" + str(hits[h]["query_length"]) + "\t")
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
195 f.write(hits[h]["accession"] + "\t" + hits[h]["description"] + "\t")
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
196 f.write(hits[h]["organism"] + "\t" + str(hits[h]["percentIdentity"]) + "\t")
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
197 f.write(str(hits[h]["num_hsps"]) + "\t" + str(hits[h]["queryOverlap"]) + "\t")
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
198 f.write(str(hits[h]["hitOverlap"]) + "\t" + str(hits[h]["evalue"]) + "\t")
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
199 f.write(str(hits[h]["score"]) + "\t" + str(hits[h]["tax_id"]) + "\t")
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
200 if h in fasta:
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
201 f.write(hits[h]["taxonomy"] + "\t" + str(fasta[h].seq))
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
202 else:
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
203 f.write(hits[h]["taxonomy"] + "\t\"\"")
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
204 f.write("\n")
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
205 else:
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
206 f.write(options.algo + "\t" + h + "\t" + read_nb + "\t" + str(hits[h])[1:-1] + "\t")
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
207 f.write("\n")
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
208 f.close()
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
209 _create_abundance(options, tsv_file)
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
210
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
211
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
212 def _create_abundance(options, tsv_file):
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
213 """
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
214 extract values from tsv files
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
215 and create abundance files
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
216 """
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
217 log.info("Calculating abundance.")
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
218 file_path = tsv_file
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
219 abundance = dict()
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
220 with open(tsv_file, 'r') as current_file:
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
221 log.debug("Reading " + file_path)
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
222 csv_reader = csv.reader(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
223 line_count = 0
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
224 for row in csv_reader:
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
225 if line_count == 0:
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
226 # headers
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
227 line_count += 1
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
228 else:
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
229 # no annotation
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
230 if len(row) == 16:
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
231 if row[14] != "":
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
232 nb_reads = row[2]
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
233 if nb_reads == "":
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
234 current_reads_nb = 0
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
235 log.debug("No reads number for " + row[1])
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
236 else:
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
237 current_reads_nb = int(nb_reads)
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
238 contig_id = row[14]
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
239 if contig_id in abundance:
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
240 # add reads
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
241 abundance[contig_id]["reads_nb"] = abundance[row[14]]["reads_nb"] + current_reads_nb
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
242 abundance[contig_id]["contigs_nb"] = abundance[row[14]]["contigs_nb"] + 1
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
243 else:
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
244 # init reads for this taxo
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
245 abundance[contig_id] = {}
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
246 abundance[contig_id]["reads_nb"] = current_reads_nb
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
247 abundance[contig_id]["contigs_nb"] = 1
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
248 else:
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
249 log.debug("No annotations for contig " + row[1])
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
250 else:
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
251 log.debug("No annotations for contig " + row[1])
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
252 log.debug(abundance)
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
253 reads_file = open(options.output + "/blast2tsv_reads.txt", "w+")
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
254 for taxo in abundance:
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
255 reads_file.write(str(abundance[taxo]["reads_nb"]))
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
256 reads_file.write("\t")
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
257 reads_file.write("\t".join(taxo.split(";")))
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
258 reads_file.write("\n")
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
259 reads_file.close()
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
260 log.info("Abundance file created " + options.output + "/blast2tsv_reads.txt")
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
261 contigs_file = open(options.output + "/blast2tsv_contigs.txt", "w+")
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
262 for taxo in abundance:
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
263 contigs_file.write(str(abundance[taxo]["contigs_nb"]))
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
264 contigs_file.write("\t")
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
265 contigs_file.write("\t".join(taxo.split(";")))
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
266 contigs_file.write("\n")
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
267 contigs_file.close()
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
268 log.info("Abundance file created " + options.output + "/blast2tsv_contigs.txt")
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
269
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
270
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
271 def _set_options():
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
272 parser = argparse.ArgumentParser()
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
273 parser.add_argument('-x', '--xml', help='XML files with results of blast', action='store', required=True, dest='xml_file')
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
274 parser.add_argument('-rn', '--read-count', help='Tab-delimited file associating seqID with read number.', action='store', dest='rn_file')
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
275 parser.add_argument('-c', '--contigs', help='FASTA file with contigs sequence.', action='store', required=True, dest='fasta_file')
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
276 parser.add_argument('-me', '--max_evalue', help='Max evalue', action='store', type=float, default=0.0001, dest='max_evalue')
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
277 parser.add_argument('-qov', '--min_query_overlap', help='Minimum query overlap', action='store', type=int, default=5, dest='min_qov')
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
278 parser.add_argument('-mhov', '--min_hit_overlap', help='Minimum hit overlap', action='store', type=int, default=5, dest='min_hov')
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
279 parser.add_argument('-s', '--min_score', help='Minimum score', action='store', type=int, default=30, dest='min_score')
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
280 parser.add_argument('-a', '--algo', help='Blast type detection (BLASTN|BLASTP|BLASTX|TBLASTX|TBLASTN|DIAMONDX).', action='store', type=str, default='BLASTX', dest='algo')
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
281 parser.add_argument('-o', '--out', help='The output file (.csv).', action='store', type=str, default='./blast2tsv', dest='output')
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
282 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
283 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
284 return args
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
285
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
286
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
287 def _set_log_level(verbosity):
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
288 if verbosity == 1:
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
289 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
290 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
291 elif verbosity == 3:
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
292 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
293 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
294
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
295
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
296 if __name__ == "__main__":
e889010415a1 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/virAnnot commit 3a3b40c15ae5e82334f016e88b1f3c5bbbb3b2cd
iuc
parents:
diff changeset
297 main()