Mercurial > repos > iuc > jbrowse

annotate blastxml_to_gapped_gff3.py @ 1:497c6bb3b717 draft

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/jbrowse commit 0887009a23d176b21536c9fd8a18c4fecc417d4f
author iuc
date Thu, 18 Jun 2015 12:10:51 -0400
parents
children 7342f467507b
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
rev   line source
1
497c6bb3b717 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/jbrowse commit 0887009a23d176b21536c9fd8a18c4fecc417d4f
iuc
parents:
diff changeset
1 #!/usr/bin/perl
497c6bb3b717 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/jbrowse commit 0887009a23d176b21536c9fd8a18c4fecc417d4f
iuc
parents:
diff changeset
2 import re
497c6bb3b717 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/jbrowse commit 0887009a23d176b21536c9fd8a18c4fecc417d4f
iuc
parents:
diff changeset
3 import sys
497c6bb3b717 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/jbrowse commit 0887009a23d176b21536c9fd8a18c4fecc417d4f
iuc
parents:
diff changeset
4 import copy
497c6bb3b717 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/jbrowse commit 0887009a23d176b21536c9fd8a18c4fecc417d4f
iuc
parents:
diff changeset
5 import argparse
497c6bb3b717 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/jbrowse commit 0887009a23d176b21536c9fd8a18c4fecc417d4f
iuc
parents:
diff changeset
6 from BCBio import GFF
497c6bb3b717 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/jbrowse commit 0887009a23d176b21536c9fd8a18c4fecc417d4f
iuc
parents:
diff changeset
7 import logging
497c6bb3b717 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/jbrowse commit 0887009a23d176b21536c9fd8a18c4fecc417d4f
iuc
parents:
diff changeset
8 logging.basicConfig(level=logging.INFO)
497c6bb3b717 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/jbrowse commit 0887009a23d176b21536c9fd8a18c4fecc417d4f
iuc
parents:
diff changeset
9 log = logging.getLogger(name='blastxml2gff3')
497c6bb3b717 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/jbrowse commit 0887009a23d176b21536c9fd8a18c4fecc417d4f
iuc
parents:
diff changeset
10
497c6bb3b717 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/jbrowse commit 0887009a23d176b21536c9fd8a18c4fecc417d4f
iuc
parents:
diff changeset
11 __author__ = "Eric Rasche"
497c6bb3b717 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/jbrowse commit 0887009a23d176b21536c9fd8a18c4fecc417d4f
iuc
parents:
diff changeset
12 __version__ = "0.4.0"
497c6bb3b717 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/jbrowse commit 0887009a23d176b21536c9fd8a18c4fecc417d4f
iuc
parents:
diff changeset
13 __maintainer__ = "Eric Rasche"
497c6bb3b717 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/jbrowse commit 0887009a23d176b21536c9fd8a18c4fecc417d4f
iuc
parents:
diff changeset
14 __email__ = "esr@tamu.edu"
497c6bb3b717 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/jbrowse commit 0887009a23d176b21536c9fd8a18c4fecc417d4f
iuc
parents:
diff changeset
15
497c6bb3b717 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/jbrowse commit 0887009a23d176b21536c9fd8a18c4fecc417d4f
iuc
parents:
diff changeset
16 __doc__ = """
497c6bb3b717 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/jbrowse commit 0887009a23d176b21536c9fd8a18c4fecc417d4f
iuc
parents:
diff changeset
17 BlastXML files, when transformed to GFF3, do not normally show gaps in the
497c6bb3b717 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/jbrowse commit 0887009a23d176b21536c9fd8a18c4fecc417d4f
iuc
parents:
diff changeset
18 blast hits. This tool aims to fill that "gap".
497c6bb3b717 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/jbrowse commit 0887009a23d176b21536c9fd8a18c4fecc417d4f
iuc
parents:
diff changeset
19 """
497c6bb3b717 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/jbrowse commit 0887009a23d176b21536c9fd8a18c4fecc417d4f
iuc
parents:
diff changeset
20
497c6bb3b717 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/jbrowse commit 0887009a23d176b21536c9fd8a18c4fecc417d4f
iuc
parents:
diff changeset
21
497c6bb3b717 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/jbrowse commit 0887009a23d176b21536c9fd8a18c4fecc417d4f
iuc
parents:
diff changeset
22 def blastxml2gff3(blastxml, min_gap=3, trim=False, trim_end=False):
497c6bb3b717 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/jbrowse commit 0887009a23d176b21536c9fd8a18c4fecc417d4f
iuc
parents:
diff changeset
23 from Bio.Blast import NCBIXML
497c6bb3b717 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/jbrowse commit 0887009a23d176b21536c9fd8a18c4fecc417d4f
iuc
parents:
diff changeset
24 from Bio.Seq import Seq
497c6bb3b717 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/jbrowse commit 0887009a23d176b21536c9fd8a18c4fecc417d4f
iuc
parents:
diff changeset
25 from Bio.SeqRecord import SeqRecord
497c6bb3b717 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/jbrowse commit 0887009a23d176b21536c9fd8a18c4fecc417d4f
iuc
parents:
diff changeset
26 from Bio.SeqFeature import SeqFeature, FeatureLocation
497c6bb3b717 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/jbrowse commit 0887009a23d176b21536c9fd8a18c4fecc417d4f
iuc
parents:
diff changeset
27
497c6bb3b717 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/jbrowse commit 0887009a23d176b21536c9fd8a18c4fecc417d4f
iuc
parents:
diff changeset
28 blast_records = NCBIXML.parse(blastxml)
497c6bb3b717 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/jbrowse commit 0887009a23d176b21536c9fd8a18c4fecc417d4f
iuc
parents:
diff changeset
29 records = []
497c6bb3b717 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/jbrowse commit 0887009a23d176b21536c9fd8a18c4fecc417d4f
iuc
parents:
diff changeset
30 for record in blast_records:
497c6bb3b717 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/jbrowse commit 0887009a23d176b21536c9fd8a18c4fecc417d4f
iuc
parents:
diff changeset
31 rec = SeqRecord(Seq("ACTG"), id=record.query)
497c6bb3b717 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/jbrowse commit 0887009a23d176b21536c9fd8a18c4fecc417d4f
iuc
parents:
diff changeset
32 for hit in record.alignments:
497c6bb3b717 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/jbrowse commit 0887009a23d176b21536c9fd8a18c4fecc417d4f
iuc
parents:
diff changeset
33 for hsp in hit.hsps:
497c6bb3b717 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/jbrowse commit 0887009a23d176b21536c9fd8a18c4fecc417d4f
iuc
parents:
diff changeset
34 qualifiers = {
497c6bb3b717 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/jbrowse commit 0887009a23d176b21536c9fd8a18c4fecc417d4f
iuc
parents:
diff changeset
35 "source": "blast",
497c6bb3b717 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/jbrowse commit 0887009a23d176b21536c9fd8a18c4fecc417d4f
iuc
parents:
diff changeset
36 "score": hsp.expect,
497c6bb3b717 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/jbrowse commit 0887009a23d176b21536c9fd8a18c4fecc417d4f
iuc
parents:
diff changeset
37 "accession": hit.accession,
497c6bb3b717 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/jbrowse commit 0887009a23d176b21536c9fd8a18c4fecc417d4f
iuc
parents:
diff changeset
38 "hit_id": hit.hit_id,
497c6bb3b717 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/jbrowse commit 0887009a23d176b21536c9fd8a18c4fecc417d4f
iuc
parents:
diff changeset
39 "length": hit.length,
497c6bb3b717 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/jbrowse commit 0887009a23d176b21536c9fd8a18c4fecc417d4f
iuc
parents:
diff changeset
40 "hit_titles": hit.title.split(' >')
497c6bb3b717 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/jbrowse commit 0887009a23d176b21536c9fd8a18c4fecc417d4f
iuc
parents:
diff changeset
41 }
497c6bb3b717 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/jbrowse commit 0887009a23d176b21536c9fd8a18c4fecc417d4f
iuc
parents:
diff changeset
42 desc = hit.title.split(' >')[0]
497c6bb3b717 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/jbrowse commit 0887009a23d176b21536c9fd8a18c4fecc417d4f
iuc
parents:
diff changeset
43 qualifiers['description'] = desc[desc.index(' '):]
497c6bb3b717 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/jbrowse commit 0887009a23d176b21536c9fd8a18c4fecc417d4f
iuc
parents:
diff changeset
44
497c6bb3b717 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/jbrowse commit 0887009a23d176b21536c9fd8a18c4fecc417d4f
iuc
parents:
diff changeset
45 # This required a fair bit of sketching out/match to figure out
497c6bb3b717 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/jbrowse commit 0887009a23d176b21536c9fd8a18c4fecc417d4f
iuc
parents:
diff changeset
46 # the first time.
497c6bb3b717 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/jbrowse commit 0887009a23d176b21536c9fd8a18c4fecc417d4f
iuc
parents:
diff changeset
47 #
497c6bb3b717 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/jbrowse commit 0887009a23d176b21536c9fd8a18c4fecc417d4f
iuc
parents:
diff changeset
48 # the match_start location must account for queries and
497c6bb3b717 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/jbrowse commit 0887009a23d176b21536c9fd8a18c4fecc417d4f
iuc
parents:
diff changeset
49 # subjecst that start at locations other than 1
497c6bb3b717 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/jbrowse commit 0887009a23d176b21536c9fd8a18c4fecc417d4f
iuc
parents:
diff changeset
50 parent_match_start = hsp.query_start - hsp.sbjct_start
497c6bb3b717 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/jbrowse commit 0887009a23d176b21536c9fd8a18c4fecc417d4f
iuc
parents:
diff changeset
51 # The end is the start + hit.length because the match itself
497c6bb3b717 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/jbrowse commit 0887009a23d176b21536c9fd8a18c4fecc417d4f
iuc
parents:
diff changeset
52 # may be longer than the parent feature, so we use the supplied
497c6bb3b717 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/jbrowse commit 0887009a23d176b21536c9fd8a18c4fecc417d4f
iuc
parents:
diff changeset
53 # subject/hit length to calculate the real ending of the target
497c6bb3b717 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/jbrowse commit 0887009a23d176b21536c9fd8a18c4fecc417d4f
iuc
parents:
diff changeset
54 # protein.
497c6bb3b717 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/jbrowse commit 0887009a23d176b21536c9fd8a18c4fecc417d4f
iuc
parents:
diff changeset
55 parent_match_end = hsp.query_start + hit.length + hsp.query.count('-')
497c6bb3b717 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/jbrowse commit 0887009a23d176b21536c9fd8a18c4fecc417d4f
iuc
parents:
diff changeset
56
497c6bb3b717 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/jbrowse commit 0887009a23d176b21536c9fd8a18c4fecc417d4f
iuc
parents:
diff changeset
57 # However, if the user requests that we trim the feature, then
497c6bb3b717 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/jbrowse commit 0887009a23d176b21536c9fd8a18c4fecc417d4f
iuc
parents:
diff changeset
58 # we need to cut the ``match`` start to 0 to match the parent feature.
497c6bb3b717 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/jbrowse commit 0887009a23d176b21536c9fd8a18c4fecc417d4f
iuc
parents:
diff changeset
59 # We'll also need to cut the end to match the query's end. It (maybe)
497c6bb3b717 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/jbrowse commit 0887009a23d176b21536c9fd8a18c4fecc417d4f
iuc
parents:
diff changeset
60 # should be the feature end? But we don't have access to that data, so
497c6bb3b717 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/jbrowse commit 0887009a23d176b21536c9fd8a18c4fecc417d4f
iuc
parents:
diff changeset
61 # We settle for this.
497c6bb3b717 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/jbrowse commit 0887009a23d176b21536c9fd8a18c4fecc417d4f
iuc
parents:
diff changeset
62 if trim:
497c6bb3b717 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/jbrowse commit 0887009a23d176b21536c9fd8a18c4fecc417d4f
iuc
parents:
diff changeset
63 if parent_match_start < 1:
497c6bb3b717 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/jbrowse commit 0887009a23d176b21536c9fd8a18c4fecc417d4f
iuc
parents:
diff changeset
64 parent_match_start = 0
497c6bb3b717 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/jbrowse commit 0887009a23d176b21536c9fd8a18c4fecc417d4f
iuc
parents:
diff changeset
65
497c6bb3b717 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/jbrowse commit 0887009a23d176b21536c9fd8a18c4fecc417d4f
iuc
parents:
diff changeset
66 if trim or trim_end:
497c6bb3b717 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/jbrowse commit 0887009a23d176b21536c9fd8a18c4fecc417d4f
iuc
parents:
diff changeset
67 if parent_match_end > hsp.query_end:
497c6bb3b717 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/jbrowse commit 0887009a23d176b21536c9fd8a18c4fecc417d4f
iuc
parents:
diff changeset
68 parent_match_end = hsp.query_end + 1
497c6bb3b717 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/jbrowse commit 0887009a23d176b21536c9fd8a18c4fecc417d4f
iuc
parents:
diff changeset
69
497c6bb3b717 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/jbrowse commit 0887009a23d176b21536c9fd8a18c4fecc417d4f
iuc
parents:
diff changeset
70 # The ``protein_match`` feature will hold one or more ``match_part``s
497c6bb3b717 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/jbrowse commit 0887009a23d176b21536c9fd8a18c4fecc417d4f
iuc
parents:
diff changeset
71 top_feature = SeqFeature(
497c6bb3b717 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/jbrowse commit 0887009a23d176b21536c9fd8a18c4fecc417d4f
iuc
parents:
diff changeset
72 FeatureLocation(parent_match_start, parent_match_end),
497c6bb3b717 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/jbrowse commit 0887009a23d176b21536c9fd8a18c4fecc417d4f
iuc
parents:
diff changeset
73 type="protein_match", strand=0,
497c6bb3b717 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/jbrowse commit 0887009a23d176b21536c9fd8a18c4fecc417d4f
iuc
parents:
diff changeset
74 qualifiers=qualifiers
497c6bb3b717 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/jbrowse commit 0887009a23d176b21536c9fd8a18c4fecc417d4f
iuc
parents:
diff changeset
75 )
497c6bb3b717 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/jbrowse commit 0887009a23d176b21536c9fd8a18c4fecc417d4f
iuc
parents:
diff changeset
76
497c6bb3b717 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/jbrowse commit 0887009a23d176b21536c9fd8a18c4fecc417d4f
iuc
parents:
diff changeset
77 # Unlike the parent feature, ``match_part``s have sources.
497c6bb3b717 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/jbrowse commit 0887009a23d176b21536c9fd8a18c4fecc417d4f
iuc
parents:
diff changeset
78 part_qualifiers = {
497c6bb3b717 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/jbrowse commit 0887009a23d176b21536c9fd8a18c4fecc417d4f
iuc
parents:
diff changeset
79 "source": "blast",
497c6bb3b717 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/jbrowse commit 0887009a23d176b21536c9fd8a18c4fecc417d4f
iuc
parents:
diff changeset
80 }
497c6bb3b717 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/jbrowse commit 0887009a23d176b21536c9fd8a18c4fecc417d4f
iuc
parents:
diff changeset
81 top_feature.sub_features = []
497c6bb3b717 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/jbrowse commit 0887009a23d176b21536c9fd8a18c4fecc417d4f
iuc
parents:
diff changeset
82 for start, end, cigar in generate_parts(hsp.query, hsp.match,
497c6bb3b717 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/jbrowse commit 0887009a23d176b21536c9fd8a18c4fecc417d4f
iuc
parents:
diff changeset
83 hsp.sbjct,
497c6bb3b717 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/jbrowse commit 0887009a23d176b21536c9fd8a18c4fecc417d4f
iuc
parents:
diff changeset
84 ignore_under=min_gap):
497c6bb3b717 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/jbrowse commit 0887009a23d176b21536c9fd8a18c4fecc417d4f
iuc
parents:
diff changeset
85 part_qualifiers['Gap'] = cigar
497c6bb3b717 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/jbrowse commit 0887009a23d176b21536c9fd8a18c4fecc417d4f
iuc
parents:
diff changeset
86 part_qualifiers['ID'] = hit.hit_id
497c6bb3b717 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/jbrowse commit 0887009a23d176b21536c9fd8a18c4fecc417d4f
iuc
parents:
diff changeset
87
497c6bb3b717 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/jbrowse commit 0887009a23d176b21536c9fd8a18c4fecc417d4f
iuc
parents:
diff changeset
88 if trim:
497c6bb3b717 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/jbrowse commit 0887009a23d176b21536c9fd8a18c4fecc417d4f
iuc
parents:
diff changeset
89 # If trimming, then we start relative to the
497c6bb3b717 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/jbrowse commit 0887009a23d176b21536c9fd8a18c4fecc417d4f
iuc
parents:
diff changeset
90 # protein_match's start
497c6bb3b717 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/jbrowse commit 0887009a23d176b21536c9fd8a18c4fecc417d4f
iuc
parents:
diff changeset
91 match_part_start = parent_match_start + start
497c6bb3b717 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/jbrowse commit 0887009a23d176b21536c9fd8a18c4fecc417d4f
iuc
parents:
diff changeset
92 else:
497c6bb3b717 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/jbrowse commit 0887009a23d176b21536c9fd8a18c4fecc417d4f
iuc
parents:
diff changeset
93 # Otherwise, we have to account for the subject start's location
497c6bb3b717 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/jbrowse commit 0887009a23d176b21536c9fd8a18c4fecc417d4f
iuc
parents:
diff changeset
94 match_part_start = parent_match_start + hsp.sbjct_start + start - 1
497c6bb3b717 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/jbrowse commit 0887009a23d176b21536c9fd8a18c4fecc417d4f
iuc
parents:
diff changeset
95
497c6bb3b717 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/jbrowse commit 0887009a23d176b21536c9fd8a18c4fecc417d4f
iuc
parents:
diff changeset
96 # We used to use hsp.align_length here, but that includes
497c6bb3b717 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/jbrowse commit 0887009a23d176b21536c9fd8a18c4fecc417d4f
iuc
parents:
diff changeset
97 # gaps in the parent sequence
497c6bb3b717 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/jbrowse commit 0887009a23d176b21536c9fd8a18c4fecc417d4f
iuc
parents:
diff changeset
98 #
497c6bb3b717 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/jbrowse commit 0887009a23d176b21536c9fd8a18c4fecc417d4f
iuc
parents:
diff changeset
99 # Furthermore align_length will give calculation errors in weird places
497c6bb3b717 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/jbrowse commit 0887009a23d176b21536c9fd8a18c4fecc417d4f
iuc
parents:
diff changeset
100 # So we just use (end-start) for simplicity
497c6bb3b717 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/jbrowse commit 0887009a23d176b21536c9fd8a18c4fecc417d4f
iuc
parents:
diff changeset
101 match_part_end = match_part_start + (end - start)
497c6bb3b717 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/jbrowse commit 0887009a23d176b21536c9fd8a18c4fecc417d4f
iuc
parents:
diff changeset
102
497c6bb3b717 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/jbrowse commit 0887009a23d176b21536c9fd8a18c4fecc417d4f
iuc
parents:
diff changeset
103 top_feature.sub_features.append(
497c6bb3b717 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/jbrowse commit 0887009a23d176b21536c9fd8a18c4fecc417d4f
iuc
parents:
diff changeset
104 SeqFeature(
497c6bb3b717 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/jbrowse commit 0887009a23d176b21536c9fd8a18c4fecc417d4f
iuc
parents:
diff changeset
105 FeatureLocation(match_part_start, match_part_end),
497c6bb3b717 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/jbrowse commit 0887009a23d176b21536c9fd8a18c4fecc417d4f
iuc
parents:
diff changeset
106 type="match_part", strand=0,
497c6bb3b717 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/jbrowse commit 0887009a23d176b21536c9fd8a18c4fecc417d4f
iuc
parents:
diff changeset
107 qualifiers=copy.deepcopy(part_qualifiers))
497c6bb3b717 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/jbrowse commit 0887009a23d176b21536c9fd8a18c4fecc417d4f
iuc
parents:
diff changeset
108 )
497c6bb3b717 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/jbrowse commit 0887009a23d176b21536c9fd8a18c4fecc417d4f
iuc
parents:
diff changeset
109
497c6bb3b717 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/jbrowse commit 0887009a23d176b21536c9fd8a18c4fecc417d4f
iuc
parents:
diff changeset
110 rec.features.append(top_feature)
497c6bb3b717 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/jbrowse commit 0887009a23d176b21536c9fd8a18c4fecc417d4f
iuc
parents:
diff changeset
111 records.append(rec)
497c6bb3b717 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/jbrowse commit 0887009a23d176b21536c9fd8a18c4fecc417d4f
iuc
parents:
diff changeset
112 return records
497c6bb3b717 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/jbrowse commit 0887009a23d176b21536c9fd8a18c4fecc417d4f
iuc
parents:
diff changeset
113
497c6bb3b717 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/jbrowse commit 0887009a23d176b21536c9fd8a18c4fecc417d4f
iuc
parents:
diff changeset
114
497c6bb3b717 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/jbrowse commit 0887009a23d176b21536c9fd8a18c4fecc417d4f
iuc
parents:
diff changeset
115 def __remove_query_gaps(query, match, subject):
497c6bb3b717 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/jbrowse commit 0887009a23d176b21536c9fd8a18c4fecc417d4f
iuc
parents:
diff changeset
116 """remove positions in all three based on gaps in query
497c6bb3b717 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/jbrowse commit 0887009a23d176b21536c9fd8a18c4fecc417d4f
iuc
parents:
diff changeset
117
497c6bb3b717 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/jbrowse commit 0887009a23d176b21536c9fd8a18c4fecc417d4f
iuc
parents:
diff changeset
118 In order to simplify math and calculations...we remove all of the gaps
497c6bb3b717 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/jbrowse commit 0887009a23d176b21536c9fd8a18c4fecc417d4f
iuc
parents:
diff changeset
119 based on gap locations in the query sequence::
497c6bb3b717 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/jbrowse commit 0887009a23d176b21536c9fd8a18c4fecc417d4f
iuc
parents:
diff changeset
120
497c6bb3b717 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/jbrowse commit 0887009a23d176b21536c9fd8a18c4fecc417d4f
iuc
parents:
diff changeset
121 Q:ACTG-ACTGACTG
497c6bb3b717 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/jbrowse commit 0887009a23d176b21536c9fd8a18c4fecc417d4f
iuc
parents:
diff changeset
122 S:ACTGAAC---CTG
497c6bb3b717 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/jbrowse commit 0887009a23d176b21536c9fd8a18c4fecc417d4f
iuc
parents:
diff changeset
123
497c6bb3b717 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/jbrowse commit 0887009a23d176b21536c9fd8a18c4fecc417d4f
iuc
parents:
diff changeset
124 will become::
497c6bb3b717 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/jbrowse commit 0887009a23d176b21536c9fd8a18c4fecc417d4f
iuc
parents:
diff changeset
125
497c6bb3b717 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/jbrowse commit 0887009a23d176b21536c9fd8a18c4fecc417d4f
iuc
parents:
diff changeset
126 Q:ACTGACTGACTG
497c6bb3b717 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/jbrowse commit 0887009a23d176b21536c9fd8a18c4fecc417d4f
iuc
parents:
diff changeset
127 S:ACTGAC---CTG
497c6bb3b717 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/jbrowse commit 0887009a23d176b21536c9fd8a18c4fecc417d4f
iuc
parents:
diff changeset
128
497c6bb3b717 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/jbrowse commit 0887009a23d176b21536c9fd8a18c4fecc417d4f
iuc
parents:
diff changeset
129 which greatly simplifies the process of identifying the correct location
497c6bb3b717 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/jbrowse commit 0887009a23d176b21536c9fd8a18c4fecc417d4f
iuc
parents:
diff changeset
130 for a match_part
497c6bb3b717 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/jbrowse commit 0887009a23d176b21536c9fd8a18c4fecc417d4f
iuc
parents:
diff changeset
131 """
497c6bb3b717 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/jbrowse commit 0887009a23d176b21536c9fd8a18c4fecc417d4f
iuc
parents:
diff changeset
132 prev = 0
497c6bb3b717 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/jbrowse commit 0887009a23d176b21536c9fd8a18c4fecc417d4f
iuc
parents:
diff changeset
133 fq = ''
497c6bb3b717 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/jbrowse commit 0887009a23d176b21536c9fd8a18c4fecc417d4f
iuc
parents:
diff changeset
134 fm = ''
497c6bb3b717 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/jbrowse commit 0887009a23d176b21536c9fd8a18c4fecc417d4f
iuc
parents:
diff changeset
135 fs = ''
497c6bb3b717 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/jbrowse commit 0887009a23d176b21536c9fd8a18c4fecc417d4f
iuc
parents:
diff changeset
136 for position in re.finditer('-', query):
497c6bb3b717 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/jbrowse commit 0887009a23d176b21536c9fd8a18c4fecc417d4f
iuc
parents:
diff changeset
137 fq += query[prev:position.start()]
497c6bb3b717 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/jbrowse commit 0887009a23d176b21536c9fd8a18c4fecc417d4f
iuc
parents:
diff changeset
138 fm += match[prev:position.start()]
497c6bb3b717 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/jbrowse commit 0887009a23d176b21536c9fd8a18c4fecc417d4f
iuc
parents:
diff changeset
139 fs += subject[prev:position.start()]
497c6bb3b717 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/jbrowse commit 0887009a23d176b21536c9fd8a18c4fecc417d4f
iuc
parents:
diff changeset
140 prev = position.start() + 1
497c6bb3b717 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/jbrowse commit 0887009a23d176b21536c9fd8a18c4fecc417d4f
iuc
parents:
diff changeset
141 fq += query[prev:]
497c6bb3b717 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/jbrowse commit 0887009a23d176b21536c9fd8a18c4fecc417d4f
iuc
parents:
diff changeset
142 fm += match[prev:]
497c6bb3b717 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/jbrowse commit 0887009a23d176b21536c9fd8a18c4fecc417d4f
iuc
parents:
diff changeset
143 fs += subject[prev:]
497c6bb3b717 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/jbrowse commit 0887009a23d176b21536c9fd8a18c4fecc417d4f
iuc
parents:
diff changeset
144
497c6bb3b717 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/jbrowse commit 0887009a23d176b21536c9fd8a18c4fecc417d4f
iuc
parents:
diff changeset
145 return (fq, fm, fs)
497c6bb3b717 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/jbrowse commit 0887009a23d176b21536c9fd8a18c4fecc417d4f
iuc
parents:
diff changeset
146
497c6bb3b717 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/jbrowse commit 0887009a23d176b21536c9fd8a18c4fecc417d4f
iuc
parents:
diff changeset
147
497c6bb3b717 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/jbrowse commit 0887009a23d176b21536c9fd8a18c4fecc417d4f
iuc
parents:
diff changeset
148 def generate_parts(query, match, subject, ignore_under=3):
497c6bb3b717 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/jbrowse commit 0887009a23d176b21536c9fd8a18c4fecc417d4f
iuc
parents:
diff changeset
149 region_q = []
497c6bb3b717 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/jbrowse commit 0887009a23d176b21536c9fd8a18c4fecc417d4f
iuc
parents:
diff changeset
150 region_m = []
497c6bb3b717 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/jbrowse commit 0887009a23d176b21536c9fd8a18c4fecc417d4f
iuc
parents:
diff changeset
151 region_s = []
497c6bb3b717 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/jbrowse commit 0887009a23d176b21536c9fd8a18c4fecc417d4f
iuc
parents:
diff changeset
152
497c6bb3b717 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/jbrowse commit 0887009a23d176b21536c9fd8a18c4fecc417d4f
iuc
parents:
diff changeset
153 (query, match, subject) = __remove_query_gaps(query, match, subject)
497c6bb3b717 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/jbrowse commit 0887009a23d176b21536c9fd8a18c4fecc417d4f
iuc
parents:
diff changeset
154
497c6bb3b717 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/jbrowse commit 0887009a23d176b21536c9fd8a18c4fecc417d4f
iuc
parents:
diff changeset
155 region_start = -1
497c6bb3b717 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/jbrowse commit 0887009a23d176b21536c9fd8a18c4fecc417d4f
iuc
parents:
diff changeset
156 region_end = -1
497c6bb3b717 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/jbrowse commit 0887009a23d176b21536c9fd8a18c4fecc417d4f
iuc
parents:
diff changeset
157 mismatch_count = 0
497c6bb3b717 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/jbrowse commit 0887009a23d176b21536c9fd8a18c4fecc417d4f
iuc
parents:
diff changeset
158 for i, (q, m, s) in enumerate(zip(query, match, subject)):
497c6bb3b717 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/jbrowse commit 0887009a23d176b21536c9fd8a18c4fecc417d4f
iuc
parents:
diff changeset
159
497c6bb3b717 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/jbrowse commit 0887009a23d176b21536c9fd8a18c4fecc417d4f
iuc
parents:
diff changeset
160 # If we have a match
497c6bb3b717 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/jbrowse commit 0887009a23d176b21536c9fd8a18c4fecc417d4f
iuc
parents:
diff changeset
161 if m != ' ' or m == '+':
497c6bb3b717 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/jbrowse commit 0887009a23d176b21536c9fd8a18c4fecc417d4f
iuc
parents:
diff changeset
162 if region_start == -1:
497c6bb3b717 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/jbrowse commit 0887009a23d176b21536c9fd8a18c4fecc417d4f
iuc
parents:
diff changeset
163 region_start = i
497c6bb3b717 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/jbrowse commit 0887009a23d176b21536c9fd8a18c4fecc417d4f
iuc
parents:
diff changeset
164 # It's a new region, we need to reset or it's pre-seeded with
497c6bb3b717 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/jbrowse commit 0887009a23d176b21536c9fd8a18c4fecc417d4f
iuc
parents:
diff changeset
165 # spaces
497c6bb3b717 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/jbrowse commit 0887009a23d176b21536c9fd8a18c4fecc417d4f
iuc
parents:
diff changeset
166 region_q = []
497c6bb3b717 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/jbrowse commit 0887009a23d176b21536c9fd8a18c4fecc417d4f
iuc
parents:
diff changeset
167 region_m = []
497c6bb3b717 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/jbrowse commit 0887009a23d176b21536c9fd8a18c4fecc417d4f
iuc
parents:
diff changeset
168 region_s = []
497c6bb3b717 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/jbrowse commit 0887009a23d176b21536c9fd8a18c4fecc417d4f
iuc
parents:
diff changeset
169 region_end = i
497c6bb3b717 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/jbrowse commit 0887009a23d176b21536c9fd8a18c4fecc417d4f
iuc
parents:
diff changeset
170 mismatch_count = 0
497c6bb3b717 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/jbrowse commit 0887009a23d176b21536c9fd8a18c4fecc417d4f
iuc
parents:
diff changeset
171 else:
497c6bb3b717 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/jbrowse commit 0887009a23d176b21536c9fd8a18c4fecc417d4f
iuc
parents:
diff changeset
172 mismatch_count += 1
497c6bb3b717 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/jbrowse commit 0887009a23d176b21536c9fd8a18c4fecc417d4f
iuc
parents:
diff changeset
173
497c6bb3b717 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/jbrowse commit 0887009a23d176b21536c9fd8a18c4fecc417d4f
iuc
parents:
diff changeset
174 region_q.append(q)
497c6bb3b717 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/jbrowse commit 0887009a23d176b21536c9fd8a18c4fecc417d4f
iuc
parents:
diff changeset
175 region_m.append(m)
497c6bb3b717 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/jbrowse commit 0887009a23d176b21536c9fd8a18c4fecc417d4f
iuc
parents:
diff changeset
176 region_s.append(s)
497c6bb3b717 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/jbrowse commit 0887009a23d176b21536c9fd8a18c4fecc417d4f
iuc
parents:
diff changeset
177
497c6bb3b717 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/jbrowse commit 0887009a23d176b21536c9fd8a18c4fecc417d4f
iuc
parents:
diff changeset
178 if mismatch_count >= ignore_under and region_start != -1 and region_end != -1:
497c6bb3b717 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/jbrowse commit 0887009a23d176b21536c9fd8a18c4fecc417d4f
iuc
parents:
diff changeset
179 region_q = region_q[0:-ignore_under]
497c6bb3b717 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/jbrowse commit 0887009a23d176b21536c9fd8a18c4fecc417d4f
iuc
parents:
diff changeset
180 region_m = region_m[0:-ignore_under]
497c6bb3b717 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/jbrowse commit 0887009a23d176b21536c9fd8a18c4fecc417d4f
iuc
parents:
diff changeset
181 region_s = region_s[0:-ignore_under]
497c6bb3b717 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/jbrowse commit 0887009a23d176b21536c9fd8a18c4fecc417d4f
iuc
parents:
diff changeset
182 yield region_start, region_end + 1, \
497c6bb3b717 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/jbrowse commit 0887009a23d176b21536c9fd8a18c4fecc417d4f
iuc
parents:
diff changeset
183 cigar_from_string(region_q, region_m, region_s, strict_m=True)
497c6bb3b717 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/jbrowse commit 0887009a23d176b21536c9fd8a18c4fecc417d4f
iuc
parents:
diff changeset
184 region_q = []
497c6bb3b717 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/jbrowse commit 0887009a23d176b21536c9fd8a18c4fecc417d4f
iuc
parents:
diff changeset
185 region_m = []
497c6bb3b717 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/jbrowse commit 0887009a23d176b21536c9fd8a18c4fecc417d4f
iuc
parents:
diff changeset
186 region_s = []
497c6bb3b717 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/jbrowse commit 0887009a23d176b21536c9fd8a18c4fecc417d4f
iuc
parents:
diff changeset
187
497c6bb3b717 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/jbrowse commit 0887009a23d176b21536c9fd8a18c4fecc417d4f
iuc
parents:
diff changeset
188 region_start = -1
497c6bb3b717 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/jbrowse commit 0887009a23d176b21536c9fd8a18c4fecc417d4f
iuc
parents:
diff changeset
189 region_end = -1
497c6bb3b717 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/jbrowse commit 0887009a23d176b21536c9fd8a18c4fecc417d4f
iuc
parents:
diff changeset
190 mismatch_count = 0
497c6bb3b717 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/jbrowse commit 0887009a23d176b21536c9fd8a18c4fecc417d4f
iuc
parents:
diff changeset
191
497c6bb3b717 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/jbrowse commit 0887009a23d176b21536c9fd8a18c4fecc417d4f
iuc
parents:
diff changeset
192 yield region_start, region_end + 1, \
497c6bb3b717 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/jbrowse commit 0887009a23d176b21536c9fd8a18c4fecc417d4f
iuc
parents:
diff changeset
193 cigar_from_string(region_q, region_m, region_s, strict_m=True)
497c6bb3b717 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/jbrowse commit 0887009a23d176b21536c9fd8a18c4fecc417d4f
iuc
parents:
diff changeset
194
497c6bb3b717 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/jbrowse commit 0887009a23d176b21536c9fd8a18c4fecc417d4f
iuc
parents:
diff changeset
195
497c6bb3b717 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/jbrowse commit 0887009a23d176b21536c9fd8a18c4fecc417d4f
iuc
parents:
diff changeset
196 def _qms_to_matches(query, match, subject, strict_m=True):
497c6bb3b717 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/jbrowse commit 0887009a23d176b21536c9fd8a18c4fecc417d4f
iuc
parents:
diff changeset
197 matchline = []
497c6bb3b717 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/jbrowse commit 0887009a23d176b21536c9fd8a18c4fecc417d4f
iuc
parents:
diff changeset
198
497c6bb3b717 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/jbrowse commit 0887009a23d176b21536c9fd8a18c4fecc417d4f
iuc
parents:
diff changeset
199 for (q, m, s) in zip(query, match, subject):
497c6bb3b717 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/jbrowse commit 0887009a23d176b21536c9fd8a18c4fecc417d4f
iuc
parents:
diff changeset
200 ret = ''
497c6bb3b717 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/jbrowse commit 0887009a23d176b21536c9fd8a18c4fecc417d4f
iuc
parents:
diff changeset
201
497c6bb3b717 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/jbrowse commit 0887009a23d176b21536c9fd8a18c4fecc417d4f
iuc
parents:
diff changeset
202 if m != ' ' or m == '+':
497c6bb3b717 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/jbrowse commit 0887009a23d176b21536c9fd8a18c4fecc417d4f
iuc
parents:
diff changeset
203 ret = '='
497c6bb3b717 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/jbrowse commit 0887009a23d176b21536c9fd8a18c4fecc417d4f
iuc
parents:
diff changeset
204 elif m == ' ':
497c6bb3b717 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/jbrowse commit 0887009a23d176b21536c9fd8a18c4fecc417d4f
iuc
parents:
diff changeset
205 if q == '-':
497c6bb3b717 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/jbrowse commit 0887009a23d176b21536c9fd8a18c4fecc417d4f
iuc
parents:
diff changeset
206 ret = 'D'
497c6bb3b717 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/jbrowse commit 0887009a23d176b21536c9fd8a18c4fecc417d4f
iuc
parents:
diff changeset
207 elif s == '-':
497c6bb3b717 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/jbrowse commit 0887009a23d176b21536c9fd8a18c4fecc417d4f
iuc
parents:
diff changeset
208 ret = 'I'
497c6bb3b717 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/jbrowse commit 0887009a23d176b21536c9fd8a18c4fecc417d4f
iuc
parents:
diff changeset
209 else:
497c6bb3b717 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/jbrowse commit 0887009a23d176b21536c9fd8a18c4fecc417d4f
iuc
parents:
diff changeset
210 ret = 'X'
497c6bb3b717 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/jbrowse commit 0887009a23d176b21536c9fd8a18c4fecc417d4f
iuc
parents:
diff changeset
211 else:
497c6bb3b717 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/jbrowse commit 0887009a23d176b21536c9fd8a18c4fecc417d4f
iuc
parents:
diff changeset
212 log.warn("Bad data: \n\t%s\n\t%s\n\t%s\n" % (query, match, subject))
497c6bb3b717 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/jbrowse commit 0887009a23d176b21536c9fd8a18c4fecc417d4f
iuc
parents:
diff changeset
213
497c6bb3b717 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/jbrowse commit 0887009a23d176b21536c9fd8a18c4fecc417d4f
iuc
parents:
diff changeset
214
497c6bb3b717 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/jbrowse commit 0887009a23d176b21536c9fd8a18c4fecc417d4f
iuc
parents:
diff changeset
215 if strict_m:
497c6bb3b717 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/jbrowse commit 0887009a23d176b21536c9fd8a18c4fecc417d4f
iuc
parents:
diff changeset
216 if ret == '=' or ret == 'X':
497c6bb3b717 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/jbrowse commit 0887009a23d176b21536c9fd8a18c4fecc417d4f
iuc
parents:
diff changeset
217 ret = 'M'
497c6bb3b717 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/jbrowse commit 0887009a23d176b21536c9fd8a18c4fecc417d4f
iuc
parents:
diff changeset
218
497c6bb3b717 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/jbrowse commit 0887009a23d176b21536c9fd8a18c4fecc417d4f
iuc
parents:
diff changeset
219 matchline.append(ret)
497c6bb3b717 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/jbrowse commit 0887009a23d176b21536c9fd8a18c4fecc417d4f
iuc
parents:
diff changeset
220 return matchline
497c6bb3b717 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/jbrowse commit 0887009a23d176b21536c9fd8a18c4fecc417d4f
iuc
parents:
diff changeset
221
497c6bb3b717 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/jbrowse commit 0887009a23d176b21536c9fd8a18c4fecc417d4f
iuc
parents:
diff changeset
222
497c6bb3b717 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/jbrowse commit 0887009a23d176b21536c9fd8a18c4fecc417d4f
iuc
parents:
diff changeset
223 def _matchline_to_cigar(matchline):
497c6bb3b717 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/jbrowse commit 0887009a23d176b21536c9fd8a18c4fecc417d4f
iuc
parents:
diff changeset
224 cigar_line = []
497c6bb3b717 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/jbrowse commit 0887009a23d176b21536c9fd8a18c4fecc417d4f
iuc
parents:
diff changeset
225 last_char = matchline[0]
497c6bb3b717 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/jbrowse commit 0887009a23d176b21536c9fd8a18c4fecc417d4f
iuc
parents:
diff changeset
226 count = 0
497c6bb3b717 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/jbrowse commit 0887009a23d176b21536c9fd8a18c4fecc417d4f
iuc
parents:
diff changeset
227 for char in matchline:
497c6bb3b717 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/jbrowse commit 0887009a23d176b21536c9fd8a18c4fecc417d4f
iuc
parents:
diff changeset
228 if char == last_char:
497c6bb3b717 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/jbrowse commit 0887009a23d176b21536c9fd8a18c4fecc417d4f
iuc
parents:
diff changeset
229 count += 1
497c6bb3b717 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/jbrowse commit 0887009a23d176b21536c9fd8a18c4fecc417d4f
iuc
parents:
diff changeset
230 else:
497c6bb3b717 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/jbrowse commit 0887009a23d176b21536c9fd8a18c4fecc417d4f
iuc
parents:
diff changeset
231 cigar_line.append("%s%s" % (last_char, count))
497c6bb3b717 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/jbrowse commit 0887009a23d176b21536c9fd8a18c4fecc417d4f
iuc
parents:
diff changeset
232 count = 1
497c6bb3b717 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/jbrowse commit 0887009a23d176b21536c9fd8a18c4fecc417d4f
iuc
parents:
diff changeset
233 last_char = char
497c6bb3b717 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/jbrowse commit 0887009a23d176b21536c9fd8a18c4fecc417d4f
iuc
parents:
diff changeset
234 cigar_line.append("%s%s" % (last_char, count))
497c6bb3b717 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/jbrowse commit 0887009a23d176b21536c9fd8a18c4fecc417d4f
iuc
parents:
diff changeset
235 return ' '.join(cigar_line)
497c6bb3b717 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/jbrowse commit 0887009a23d176b21536c9fd8a18c4fecc417d4f
iuc
parents:
diff changeset
236
497c6bb3b717 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/jbrowse commit 0887009a23d176b21536c9fd8a18c4fecc417d4f
iuc
parents:
diff changeset
237
497c6bb3b717 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/jbrowse commit 0887009a23d176b21536c9fd8a18c4fecc417d4f
iuc
parents:
diff changeset
238 def cigar_from_string(query, match, subject, strict_m=True):
497c6bb3b717 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/jbrowse commit 0887009a23d176b21536c9fd8a18c4fecc417d4f
iuc
parents:
diff changeset
239 matchline = _qms_to_matches(query, match, subject, strict_m=strict_m)
497c6bb3b717 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/jbrowse commit 0887009a23d176b21536c9fd8a18c4fecc417d4f
iuc
parents:
diff changeset
240 if len(matchline) > 0:
497c6bb3b717 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/jbrowse commit 0887009a23d176b21536c9fd8a18c4fecc417d4f
iuc
parents:
diff changeset
241 return _matchline_to_cigar(matchline)
497c6bb3b717 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/jbrowse commit 0887009a23d176b21536c9fd8a18c4fecc417d4f
iuc
parents:
diff changeset
242 else:
497c6bb3b717 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/jbrowse commit 0887009a23d176b21536c9fd8a18c4fecc417d4f
iuc
parents:
diff changeset
243 return ""
497c6bb3b717 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/jbrowse commit 0887009a23d176b21536c9fd8a18c4fecc417d4f
iuc
parents:
diff changeset
244
497c6bb3b717 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/jbrowse commit 0887009a23d176b21536c9fd8a18c4fecc417d4f
iuc
parents:
diff changeset
245
497c6bb3b717 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/jbrowse commit 0887009a23d176b21536c9fd8a18c4fecc417d4f
iuc
parents:
diff changeset
246 if __name__ == '__main__':
497c6bb3b717 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/jbrowse commit 0887009a23d176b21536c9fd8a18c4fecc417d4f
iuc
parents:
diff changeset
247 parser = argparse.ArgumentParser(description='Convert Blast XML to gapped GFF3', epilog='')
497c6bb3b717 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/jbrowse commit 0887009a23d176b21536c9fd8a18c4fecc417d4f
iuc
parents:
diff changeset
248 parser.add_argument('blastxml', type=file, help='Blast XML Output')
497c6bb3b717 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/jbrowse commit 0887009a23d176b21536c9fd8a18c4fecc417d4f
iuc
parents:
diff changeset
249 parser.add_argument('--min_gap', type=int, help='Maximum gap size before generating a new match_part', default=3)
497c6bb3b717 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/jbrowse commit 0887009a23d176b21536c9fd8a18c4fecc417d4f
iuc
parents:
diff changeset
250 parser.add_argument('--trim', action='store_true', help='Trim blast hits to be only as long as the parent feature')
497c6bb3b717 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/jbrowse commit 0887009a23d176b21536c9fd8a18c4fecc417d4f
iuc
parents:
diff changeset
251 parser.add_argument('--trim_end', action='store_true', help='Cut blast results off at end of gene')
497c6bb3b717 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/jbrowse commit 0887009a23d176b21536c9fd8a18c4fecc417d4f
iuc
parents:
diff changeset
252 args = parser.parse_args()
497c6bb3b717 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/jbrowse commit 0887009a23d176b21536c9fd8a18c4fecc417d4f
iuc
parents:
diff changeset
253
497c6bb3b717 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/jbrowse commit 0887009a23d176b21536c9fd8a18c4fecc417d4f
iuc
parents:
diff changeset
254 result = blastxml2gff3(**vars(args))
497c6bb3b717 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/jbrowse commit 0887009a23d176b21536c9fd8a18c4fecc417d4f
iuc
parents:
diff changeset
255
497c6bb3b717 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/jbrowse commit 0887009a23d176b21536c9fd8a18c4fecc417d4f
iuc
parents:
diff changeset
256 GFF.write(result, sys.stdout)