Mercurial > repos > iuc > blastxml_to_gapped_gff3
annotate blastxml_to_gapped_gff3.py @ 0:bd47051afe98 draft
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/blastxml_to_gapped_gff3 commit 8f38145c94ecb1e23c3ff6f0243213dc49d2287e
| author | iuc | 
|---|---|
| date | Tue, 20 Dec 2016 09:21:11 -0500 | 
| parents | |
| children | 877cd0833221 | 
| rev | line source | 
|---|---|
| 0 
bd47051afe98
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/blastxml_to_gapped_gff3 commit 8f38145c94ecb1e23c3ff6f0243213dc49d2287e
 iuc parents: diff
changeset | 1 #!/usr/bin/perl | 
| 
bd47051afe98
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/blastxml_to_gapped_gff3 commit 8f38145c94ecb1e23c3ff6f0243213dc49d2287e
 iuc parents: diff
changeset | 2 import argparse | 
| 
bd47051afe98
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/blastxml_to_gapped_gff3 commit 8f38145c94ecb1e23c3ff6f0243213dc49d2287e
 iuc parents: diff
changeset | 3 import copy | 
| 
bd47051afe98
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/blastxml_to_gapped_gff3 commit 8f38145c94ecb1e23c3ff6f0243213dc49d2287e
 iuc parents: diff
changeset | 4 import logging | 
| 
bd47051afe98
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/blastxml_to_gapped_gff3 commit 8f38145c94ecb1e23c3ff6f0243213dc49d2287e
 iuc parents: diff
changeset | 5 import re | 
| 
bd47051afe98
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/blastxml_to_gapped_gff3 commit 8f38145c94ecb1e23c3ff6f0243213dc49d2287e
 iuc parents: diff
changeset | 6 import sys | 
| 
bd47051afe98
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/blastxml_to_gapped_gff3 commit 8f38145c94ecb1e23c3ff6f0243213dc49d2287e
 iuc parents: diff
changeset | 7 | 
| 
bd47051afe98
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/blastxml_to_gapped_gff3 commit 8f38145c94ecb1e23c3ff6f0243213dc49d2287e
 iuc parents: diff
changeset | 8 from BCBio import GFF | 
| 
bd47051afe98
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/blastxml_to_gapped_gff3 commit 8f38145c94ecb1e23c3ff6f0243213dc49d2287e
 iuc parents: diff
changeset | 9 | 
| 
bd47051afe98
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/blastxml_to_gapped_gff3 commit 8f38145c94ecb1e23c3ff6f0243213dc49d2287e
 iuc parents: diff
changeset | 10 logging.basicConfig(level=logging.INFO) | 
| 
bd47051afe98
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/blastxml_to_gapped_gff3 commit 8f38145c94ecb1e23c3ff6f0243213dc49d2287e
 iuc parents: diff
changeset | 11 log = logging.getLogger(name='blastxml2gff3') | 
| 
bd47051afe98
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/blastxml_to_gapped_gff3 commit 8f38145c94ecb1e23c3ff6f0243213dc49d2287e
 iuc parents: diff
changeset | 12 | 
| 
bd47051afe98
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/blastxml_to_gapped_gff3 commit 8f38145c94ecb1e23c3ff6f0243213dc49d2287e
 iuc parents: diff
changeset | 13 __author__ = "Eric Rasche" | 
| 
bd47051afe98
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/blastxml_to_gapped_gff3 commit 8f38145c94ecb1e23c3ff6f0243213dc49d2287e
 iuc parents: diff
changeset | 14 __version__ = "0.4.0" | 
| 
bd47051afe98
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/blastxml_to_gapped_gff3 commit 8f38145c94ecb1e23c3ff6f0243213dc49d2287e
 iuc parents: diff
changeset | 15 __maintainer__ = "Eric Rasche" | 
| 
bd47051afe98
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/blastxml_to_gapped_gff3 commit 8f38145c94ecb1e23c3ff6f0243213dc49d2287e
 iuc parents: diff
changeset | 16 __email__ = "esr@tamu.edu" | 
| 
bd47051afe98
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/blastxml_to_gapped_gff3 commit 8f38145c94ecb1e23c3ff6f0243213dc49d2287e
 iuc parents: diff
changeset | 17 | 
| 
bd47051afe98
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/blastxml_to_gapped_gff3 commit 8f38145c94ecb1e23c3ff6f0243213dc49d2287e
 iuc parents: diff
changeset | 18 __doc__ = """ | 
| 
bd47051afe98
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/blastxml_to_gapped_gff3 commit 8f38145c94ecb1e23c3ff6f0243213dc49d2287e
 iuc parents: diff
changeset | 19 BlastXML files, when transformed to GFF3, do not normally show gaps in the | 
| 
bd47051afe98
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/blastxml_to_gapped_gff3 commit 8f38145c94ecb1e23c3ff6f0243213dc49d2287e
 iuc parents: diff
changeset | 20 blast hits. This tool aims to fill that "gap". | 
| 
bd47051afe98
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/blastxml_to_gapped_gff3 commit 8f38145c94ecb1e23c3ff6f0243213dc49d2287e
 iuc parents: diff
changeset | 21 """ | 
| 
bd47051afe98
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/blastxml_to_gapped_gff3 commit 8f38145c94ecb1e23c3ff6f0243213dc49d2287e
 iuc parents: diff
changeset | 22 | 
| 
bd47051afe98
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/blastxml_to_gapped_gff3 commit 8f38145c94ecb1e23c3ff6f0243213dc49d2287e
 iuc parents: diff
changeset | 23 | 
| 
bd47051afe98
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/blastxml_to_gapped_gff3 commit 8f38145c94ecb1e23c3ff6f0243213dc49d2287e
 iuc parents: diff
changeset | 24 def blastxml2gff3(blastxml, min_gap=3, trim=False, trim_end=False): | 
| 
bd47051afe98
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/blastxml_to_gapped_gff3 commit 8f38145c94ecb1e23c3ff6f0243213dc49d2287e
 iuc parents: diff
changeset | 25 from Bio.Blast import NCBIXML | 
| 
bd47051afe98
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/blastxml_to_gapped_gff3 commit 8f38145c94ecb1e23c3ff6f0243213dc49d2287e
 iuc parents: diff
changeset | 26 from Bio.Seq import Seq | 
| 
bd47051afe98
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/blastxml_to_gapped_gff3 commit 8f38145c94ecb1e23c3ff6f0243213dc49d2287e
 iuc parents: diff
changeset | 27 from Bio.SeqRecord import SeqRecord | 
| 
bd47051afe98
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/blastxml_to_gapped_gff3 commit 8f38145c94ecb1e23c3ff6f0243213dc49d2287e
 iuc parents: diff
changeset | 28 from Bio.SeqFeature import SeqFeature, FeatureLocation | 
| 
bd47051afe98
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/blastxml_to_gapped_gff3 commit 8f38145c94ecb1e23c3ff6f0243213dc49d2287e
 iuc parents: diff
changeset | 29 | 
| 
bd47051afe98
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/blastxml_to_gapped_gff3 commit 8f38145c94ecb1e23c3ff6f0243213dc49d2287e
 iuc parents: diff
changeset | 30 blast_records = NCBIXML.parse(blastxml) | 
| 
bd47051afe98
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/blastxml_to_gapped_gff3 commit 8f38145c94ecb1e23c3ff6f0243213dc49d2287e
 iuc parents: diff
changeset | 31 records = [] | 
| 
bd47051afe98
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/blastxml_to_gapped_gff3 commit 8f38145c94ecb1e23c3ff6f0243213dc49d2287e
 iuc parents: diff
changeset | 32 for record in blast_records: | 
| 
bd47051afe98
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/blastxml_to_gapped_gff3 commit 8f38145c94ecb1e23c3ff6f0243213dc49d2287e
 iuc parents: diff
changeset | 33 # http://www.sequenceontology.org/browser/release_2.4/term/SO:0000343 | 
| 
bd47051afe98
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/blastxml_to_gapped_gff3 commit 8f38145c94ecb1e23c3ff6f0243213dc49d2287e
 iuc parents: diff
changeset | 34 match_type = { # Currently we can only handle BLASTN, BLASTP | 
| 
bd47051afe98
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/blastxml_to_gapped_gff3 commit 8f38145c94ecb1e23c3ff6f0243213dc49d2287e
 iuc parents: diff
changeset | 35 'BLASTN': 'nucleotide_match', | 
| 
bd47051afe98
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/blastxml_to_gapped_gff3 commit 8f38145c94ecb1e23c3ff6f0243213dc49d2287e
 iuc parents: diff
changeset | 36 'BLASTP': 'protein_match', | 
| 
bd47051afe98
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/blastxml_to_gapped_gff3 commit 8f38145c94ecb1e23c3ff6f0243213dc49d2287e
 iuc parents: diff
changeset | 37 }.get(record.application, 'match') | 
| 
bd47051afe98
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/blastxml_to_gapped_gff3 commit 8f38145c94ecb1e23c3ff6f0243213dc49d2287e
 iuc parents: diff
changeset | 38 | 
| 
bd47051afe98
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/blastxml_to_gapped_gff3 commit 8f38145c94ecb1e23c3ff6f0243213dc49d2287e
 iuc parents: diff
changeset | 39 rec = SeqRecord(Seq("ACTG"), id=record.query) | 
| 
bd47051afe98
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/blastxml_to_gapped_gff3 commit 8f38145c94ecb1e23c3ff6f0243213dc49d2287e
 iuc parents: diff
changeset | 40 for hit in record.alignments: | 
| 
bd47051afe98
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/blastxml_to_gapped_gff3 commit 8f38145c94ecb1e23c3ff6f0243213dc49d2287e
 iuc parents: diff
changeset | 41 for hsp in hit.hsps: | 
| 
bd47051afe98
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/blastxml_to_gapped_gff3 commit 8f38145c94ecb1e23c3ff6f0243213dc49d2287e
 iuc parents: diff
changeset | 42 qualifiers = { | 
| 
bd47051afe98
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/blastxml_to_gapped_gff3 commit 8f38145c94ecb1e23c3ff6f0243213dc49d2287e
 iuc parents: diff
changeset | 43 "source": "blast", | 
| 
bd47051afe98
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/blastxml_to_gapped_gff3 commit 8f38145c94ecb1e23c3ff6f0243213dc49d2287e
 iuc parents: diff
changeset | 44 "score": hsp.expect, | 
| 
bd47051afe98
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/blastxml_to_gapped_gff3 commit 8f38145c94ecb1e23c3ff6f0243213dc49d2287e
 iuc parents: diff
changeset | 45 "accession": hit.accession, | 
| 
bd47051afe98
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/blastxml_to_gapped_gff3 commit 8f38145c94ecb1e23c3ff6f0243213dc49d2287e
 iuc parents: diff
changeset | 46 "hit_id": hit.hit_id, | 
| 
bd47051afe98
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/blastxml_to_gapped_gff3 commit 8f38145c94ecb1e23c3ff6f0243213dc49d2287e
 iuc parents: diff
changeset | 47 "length": hit.length, | 
| 
bd47051afe98
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/blastxml_to_gapped_gff3 commit 8f38145c94ecb1e23c3ff6f0243213dc49d2287e
 iuc parents: diff
changeset | 48 "hit_titles": hit.title.split(' >') | 
| 
bd47051afe98
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/blastxml_to_gapped_gff3 commit 8f38145c94ecb1e23c3ff6f0243213dc49d2287e
 iuc parents: diff
changeset | 49 } | 
| 
bd47051afe98
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/blastxml_to_gapped_gff3 commit 8f38145c94ecb1e23c3ff6f0243213dc49d2287e
 iuc parents: diff
changeset | 50 desc = hit.title.split(' >')[0] | 
| 
bd47051afe98
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/blastxml_to_gapped_gff3 commit 8f38145c94ecb1e23c3ff6f0243213dc49d2287e
 iuc parents: diff
changeset | 51 qualifiers['description'] = desc[desc.index(' '):] | 
| 
bd47051afe98
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/blastxml_to_gapped_gff3 commit 8f38145c94ecb1e23c3ff6f0243213dc49d2287e
 iuc parents: diff
changeset | 52 | 
| 
bd47051afe98
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/blastxml_to_gapped_gff3 commit 8f38145c94ecb1e23c3ff6f0243213dc49d2287e
 iuc parents: diff
changeset | 53 # This required a fair bit of sketching out/match to figure out | 
| 
bd47051afe98
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/blastxml_to_gapped_gff3 commit 8f38145c94ecb1e23c3ff6f0243213dc49d2287e
 iuc parents: diff
changeset | 54 # the first time. | 
| 
bd47051afe98
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/blastxml_to_gapped_gff3 commit 8f38145c94ecb1e23c3ff6f0243213dc49d2287e
 iuc parents: diff
changeset | 55 # | 
| 
bd47051afe98
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/blastxml_to_gapped_gff3 commit 8f38145c94ecb1e23c3ff6f0243213dc49d2287e
 iuc parents: diff
changeset | 56 # the match_start location must account for queries and | 
| 
bd47051afe98
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/blastxml_to_gapped_gff3 commit 8f38145c94ecb1e23c3ff6f0243213dc49d2287e
 iuc parents: diff
changeset | 57 # subjecst that start at locations other than 1 | 
| 
bd47051afe98
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/blastxml_to_gapped_gff3 commit 8f38145c94ecb1e23c3ff6f0243213dc49d2287e
 iuc parents: diff
changeset | 58 parent_match_start = hsp.query_start - hsp.sbjct_start | 
| 
bd47051afe98
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/blastxml_to_gapped_gff3 commit 8f38145c94ecb1e23c3ff6f0243213dc49d2287e
 iuc parents: diff
changeset | 59 # The end is the start + hit.length because the match itself | 
| 
bd47051afe98
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/blastxml_to_gapped_gff3 commit 8f38145c94ecb1e23c3ff6f0243213dc49d2287e
 iuc parents: diff
changeset | 60 # may be longer than the parent feature, so we use the supplied | 
| 
bd47051afe98
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/blastxml_to_gapped_gff3 commit 8f38145c94ecb1e23c3ff6f0243213dc49d2287e
 iuc parents: diff
changeset | 61 # subject/hit length to calculate the real ending of the target | 
| 
bd47051afe98
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/blastxml_to_gapped_gff3 commit 8f38145c94ecb1e23c3ff6f0243213dc49d2287e
 iuc parents: diff
changeset | 62 # protein. | 
| 
bd47051afe98
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/blastxml_to_gapped_gff3 commit 8f38145c94ecb1e23c3ff6f0243213dc49d2287e
 iuc parents: diff
changeset | 63 parent_match_end = hsp.query_start + hit.length + hsp.query.count('-') | 
| 
bd47051afe98
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/blastxml_to_gapped_gff3 commit 8f38145c94ecb1e23c3ff6f0243213dc49d2287e
 iuc parents: diff
changeset | 64 | 
| 
bd47051afe98
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/blastxml_to_gapped_gff3 commit 8f38145c94ecb1e23c3ff6f0243213dc49d2287e
 iuc parents: diff
changeset | 65 # However, if the user requests that we trim the feature, then | 
| 
bd47051afe98
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/blastxml_to_gapped_gff3 commit 8f38145c94ecb1e23c3ff6f0243213dc49d2287e
 iuc parents: diff
changeset | 66 # we need to cut the ``match`` start to 0 to match the parent feature. | 
| 
bd47051afe98
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/blastxml_to_gapped_gff3 commit 8f38145c94ecb1e23c3ff6f0243213dc49d2287e
 iuc parents: diff
changeset | 67 # We'll also need to cut the end to match the query's end. It (maybe) | 
| 
bd47051afe98
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/blastxml_to_gapped_gff3 commit 8f38145c94ecb1e23c3ff6f0243213dc49d2287e
 iuc parents: diff
changeset | 68 # should be the feature end? But we don't have access to that data, so | 
| 
bd47051afe98
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/blastxml_to_gapped_gff3 commit 8f38145c94ecb1e23c3ff6f0243213dc49d2287e
 iuc parents: diff
changeset | 69 # We settle for this. | 
| 
bd47051afe98
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/blastxml_to_gapped_gff3 commit 8f38145c94ecb1e23c3ff6f0243213dc49d2287e
 iuc parents: diff
changeset | 70 if trim: | 
| 
bd47051afe98
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/blastxml_to_gapped_gff3 commit 8f38145c94ecb1e23c3ff6f0243213dc49d2287e
 iuc parents: diff
changeset | 71 if parent_match_start < 1: | 
| 
bd47051afe98
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/blastxml_to_gapped_gff3 commit 8f38145c94ecb1e23c3ff6f0243213dc49d2287e
 iuc parents: diff
changeset | 72 parent_match_start = 0 | 
| 
bd47051afe98
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/blastxml_to_gapped_gff3 commit 8f38145c94ecb1e23c3ff6f0243213dc49d2287e
 iuc parents: diff
changeset | 73 | 
| 
bd47051afe98
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/blastxml_to_gapped_gff3 commit 8f38145c94ecb1e23c3ff6f0243213dc49d2287e
 iuc parents: diff
changeset | 74 if trim or trim_end: | 
| 
bd47051afe98
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/blastxml_to_gapped_gff3 commit 8f38145c94ecb1e23c3ff6f0243213dc49d2287e
 iuc parents: diff
changeset | 75 if parent_match_end > hsp.query_end: | 
| 
bd47051afe98
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/blastxml_to_gapped_gff3 commit 8f38145c94ecb1e23c3ff6f0243213dc49d2287e
 iuc parents: diff
changeset | 76 parent_match_end = hsp.query_end + 1 | 
| 
bd47051afe98
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/blastxml_to_gapped_gff3 commit 8f38145c94ecb1e23c3ff6f0243213dc49d2287e
 iuc parents: diff
changeset | 77 | 
| 
bd47051afe98
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/blastxml_to_gapped_gff3 commit 8f38145c94ecb1e23c3ff6f0243213dc49d2287e
 iuc parents: diff
changeset | 78 # The ``match`` feature will hold one or more ``match_part``s | 
| 
bd47051afe98
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/blastxml_to_gapped_gff3 commit 8f38145c94ecb1e23c3ff6f0243213dc49d2287e
 iuc parents: diff
changeset | 79 top_feature = SeqFeature( | 
| 
bd47051afe98
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/blastxml_to_gapped_gff3 commit 8f38145c94ecb1e23c3ff6f0243213dc49d2287e
 iuc parents: diff
changeset | 80 FeatureLocation(parent_match_start, parent_match_end), | 
| 
bd47051afe98
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/blastxml_to_gapped_gff3 commit 8f38145c94ecb1e23c3ff6f0243213dc49d2287e
 iuc parents: diff
changeset | 81 type=match_type, strand=0, | 
| 
bd47051afe98
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/blastxml_to_gapped_gff3 commit 8f38145c94ecb1e23c3ff6f0243213dc49d2287e
 iuc parents: diff
changeset | 82 qualifiers=qualifiers | 
| 
bd47051afe98
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/blastxml_to_gapped_gff3 commit 8f38145c94ecb1e23c3ff6f0243213dc49d2287e
 iuc parents: diff
changeset | 83 ) | 
| 
bd47051afe98
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/blastxml_to_gapped_gff3 commit 8f38145c94ecb1e23c3ff6f0243213dc49d2287e
 iuc parents: diff
changeset | 84 | 
| 
bd47051afe98
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/blastxml_to_gapped_gff3 commit 8f38145c94ecb1e23c3ff6f0243213dc49d2287e
 iuc parents: diff
changeset | 85 # Unlike the parent feature, ``match_part``s have sources. | 
| 
bd47051afe98
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/blastxml_to_gapped_gff3 commit 8f38145c94ecb1e23c3ff6f0243213dc49d2287e
 iuc parents: diff
changeset | 86 part_qualifiers = { | 
| 
bd47051afe98
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/blastxml_to_gapped_gff3 commit 8f38145c94ecb1e23c3ff6f0243213dc49d2287e
 iuc parents: diff
changeset | 87 "source": "blast", | 
| 
bd47051afe98
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/blastxml_to_gapped_gff3 commit 8f38145c94ecb1e23c3ff6f0243213dc49d2287e
 iuc parents: diff
changeset | 88 } | 
| 
bd47051afe98
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/blastxml_to_gapped_gff3 commit 8f38145c94ecb1e23c3ff6f0243213dc49d2287e
 iuc parents: diff
changeset | 89 top_feature.sub_features = [] | 
| 
bd47051afe98
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/blastxml_to_gapped_gff3 commit 8f38145c94ecb1e23c3ff6f0243213dc49d2287e
 iuc parents: diff
changeset | 90 for start, end, cigar in generate_parts(hsp.query, hsp.match, | 
| 
bd47051afe98
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/blastxml_to_gapped_gff3 commit 8f38145c94ecb1e23c3ff6f0243213dc49d2287e
 iuc parents: diff
changeset | 91 hsp.sbjct, | 
| 
bd47051afe98
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/blastxml_to_gapped_gff3 commit 8f38145c94ecb1e23c3ff6f0243213dc49d2287e
 iuc parents: diff
changeset | 92 ignore_under=min_gap): | 
| 
bd47051afe98
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/blastxml_to_gapped_gff3 commit 8f38145c94ecb1e23c3ff6f0243213dc49d2287e
 iuc parents: diff
changeset | 93 part_qualifiers['Gap'] = cigar | 
| 
bd47051afe98
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/blastxml_to_gapped_gff3 commit 8f38145c94ecb1e23c3ff6f0243213dc49d2287e
 iuc parents: diff
changeset | 94 part_qualifiers['ID'] = hit.hit_id | 
| 
bd47051afe98
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/blastxml_to_gapped_gff3 commit 8f38145c94ecb1e23c3ff6f0243213dc49d2287e
 iuc parents: diff
changeset | 95 | 
| 
bd47051afe98
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/blastxml_to_gapped_gff3 commit 8f38145c94ecb1e23c3ff6f0243213dc49d2287e
 iuc parents: diff
changeset | 96 if trim: | 
| 
bd47051afe98
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/blastxml_to_gapped_gff3 commit 8f38145c94ecb1e23c3ff6f0243213dc49d2287e
 iuc parents: diff
changeset | 97 # If trimming, then we start relative to the | 
| 
bd47051afe98
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/blastxml_to_gapped_gff3 commit 8f38145c94ecb1e23c3ff6f0243213dc49d2287e
 iuc parents: diff
changeset | 98 # match's start | 
| 
bd47051afe98
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/blastxml_to_gapped_gff3 commit 8f38145c94ecb1e23c3ff6f0243213dc49d2287e
 iuc parents: diff
changeset | 99 match_part_start = parent_match_start + start | 
| 
bd47051afe98
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/blastxml_to_gapped_gff3 commit 8f38145c94ecb1e23c3ff6f0243213dc49d2287e
 iuc parents: diff
changeset | 100 else: | 
| 
bd47051afe98
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/blastxml_to_gapped_gff3 commit 8f38145c94ecb1e23c3ff6f0243213dc49d2287e
 iuc parents: diff
changeset | 101 # Otherwise, we have to account for the subject start's location | 
| 
bd47051afe98
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/blastxml_to_gapped_gff3 commit 8f38145c94ecb1e23c3ff6f0243213dc49d2287e
 iuc parents: diff
changeset | 102 match_part_start = parent_match_start + hsp.sbjct_start + start - 1 | 
| 
bd47051afe98
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/blastxml_to_gapped_gff3 commit 8f38145c94ecb1e23c3ff6f0243213dc49d2287e
 iuc parents: diff
changeset | 103 | 
| 
bd47051afe98
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/blastxml_to_gapped_gff3 commit 8f38145c94ecb1e23c3ff6f0243213dc49d2287e
 iuc parents: diff
changeset | 104 # We used to use hsp.align_length here, but that includes | 
| 
bd47051afe98
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/blastxml_to_gapped_gff3 commit 8f38145c94ecb1e23c3ff6f0243213dc49d2287e
 iuc parents: diff
changeset | 105 # gaps in the parent sequence | 
| 
bd47051afe98
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/blastxml_to_gapped_gff3 commit 8f38145c94ecb1e23c3ff6f0243213dc49d2287e
 iuc parents: diff
changeset | 106 # | 
| 
bd47051afe98
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/blastxml_to_gapped_gff3 commit 8f38145c94ecb1e23c3ff6f0243213dc49d2287e
 iuc parents: diff
changeset | 107 # Furthermore align_length will give calculation errors in weird places | 
| 
bd47051afe98
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/blastxml_to_gapped_gff3 commit 8f38145c94ecb1e23c3ff6f0243213dc49d2287e
 iuc parents: diff
changeset | 108 # So we just use (end-start) for simplicity | 
| 
bd47051afe98
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/blastxml_to_gapped_gff3 commit 8f38145c94ecb1e23c3ff6f0243213dc49d2287e
 iuc parents: diff
changeset | 109 match_part_end = match_part_start + (end - start) | 
| 
bd47051afe98
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/blastxml_to_gapped_gff3 commit 8f38145c94ecb1e23c3ff6f0243213dc49d2287e
 iuc parents: diff
changeset | 110 | 
| 
bd47051afe98
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/blastxml_to_gapped_gff3 commit 8f38145c94ecb1e23c3ff6f0243213dc49d2287e
 iuc parents: diff
changeset | 111 top_feature.sub_features.append( | 
| 
bd47051afe98
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/blastxml_to_gapped_gff3 commit 8f38145c94ecb1e23c3ff6f0243213dc49d2287e
 iuc parents: diff
changeset | 112 SeqFeature( | 
| 
bd47051afe98
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/blastxml_to_gapped_gff3 commit 8f38145c94ecb1e23c3ff6f0243213dc49d2287e
 iuc parents: diff
changeset | 113 FeatureLocation(match_part_start, match_part_end), | 
| 
bd47051afe98
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/blastxml_to_gapped_gff3 commit 8f38145c94ecb1e23c3ff6f0243213dc49d2287e
 iuc parents: diff
changeset | 114 type="match_part", strand=0, | 
| 
bd47051afe98
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/blastxml_to_gapped_gff3 commit 8f38145c94ecb1e23c3ff6f0243213dc49d2287e
 iuc parents: diff
changeset | 115 qualifiers=copy.deepcopy(part_qualifiers)) | 
| 
bd47051afe98
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/blastxml_to_gapped_gff3 commit 8f38145c94ecb1e23c3ff6f0243213dc49d2287e
 iuc parents: diff
changeset | 116 ) | 
| 
bd47051afe98
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/blastxml_to_gapped_gff3 commit 8f38145c94ecb1e23c3ff6f0243213dc49d2287e
 iuc parents: diff
changeset | 117 | 
| 
bd47051afe98
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/blastxml_to_gapped_gff3 commit 8f38145c94ecb1e23c3ff6f0243213dc49d2287e
 iuc parents: diff
changeset | 118 rec.features.append(top_feature) | 
| 
bd47051afe98
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/blastxml_to_gapped_gff3 commit 8f38145c94ecb1e23c3ff6f0243213dc49d2287e
 iuc parents: diff
changeset | 119 rec.annotations = {} | 
| 
bd47051afe98
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/blastxml_to_gapped_gff3 commit 8f38145c94ecb1e23c3ff6f0243213dc49d2287e
 iuc parents: diff
changeset | 120 records.append(rec) | 
| 
bd47051afe98
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/blastxml_to_gapped_gff3 commit 8f38145c94ecb1e23c3ff6f0243213dc49d2287e
 iuc parents: diff
changeset | 121 return records | 
| 
bd47051afe98
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/blastxml_to_gapped_gff3 commit 8f38145c94ecb1e23c3ff6f0243213dc49d2287e
 iuc parents: diff
changeset | 122 | 
| 
bd47051afe98
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/blastxml_to_gapped_gff3 commit 8f38145c94ecb1e23c3ff6f0243213dc49d2287e
 iuc parents: diff
changeset | 123 | 
| 
bd47051afe98
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/blastxml_to_gapped_gff3 commit 8f38145c94ecb1e23c3ff6f0243213dc49d2287e
 iuc parents: diff
changeset | 124 def __remove_query_gaps(query, match, subject): | 
| 
bd47051afe98
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/blastxml_to_gapped_gff3 commit 8f38145c94ecb1e23c3ff6f0243213dc49d2287e
 iuc parents: diff
changeset | 125 """remove positions in all three based on gaps in query | 
| 
bd47051afe98
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/blastxml_to_gapped_gff3 commit 8f38145c94ecb1e23c3ff6f0243213dc49d2287e
 iuc parents: diff
changeset | 126 | 
| 
bd47051afe98
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/blastxml_to_gapped_gff3 commit 8f38145c94ecb1e23c3ff6f0243213dc49d2287e
 iuc parents: diff
changeset | 127 In order to simplify math and calculations...we remove all of the gaps | 
| 
bd47051afe98
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/blastxml_to_gapped_gff3 commit 8f38145c94ecb1e23c3ff6f0243213dc49d2287e
 iuc parents: diff
changeset | 128 based on gap locations in the query sequence:: | 
| 
bd47051afe98
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/blastxml_to_gapped_gff3 commit 8f38145c94ecb1e23c3ff6f0243213dc49d2287e
 iuc parents: diff
changeset | 129 | 
| 
bd47051afe98
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/blastxml_to_gapped_gff3 commit 8f38145c94ecb1e23c3ff6f0243213dc49d2287e
 iuc parents: diff
changeset | 130 Q:ACTG-ACTGACTG | 
| 
bd47051afe98
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/blastxml_to_gapped_gff3 commit 8f38145c94ecb1e23c3ff6f0243213dc49d2287e
 iuc parents: diff
changeset | 131 S:ACTGAAC---CTG | 
| 
bd47051afe98
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/blastxml_to_gapped_gff3 commit 8f38145c94ecb1e23c3ff6f0243213dc49d2287e
 iuc parents: diff
changeset | 132 | 
| 
bd47051afe98
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/blastxml_to_gapped_gff3 commit 8f38145c94ecb1e23c3ff6f0243213dc49d2287e
 iuc parents: diff
changeset | 133 will become:: | 
| 
bd47051afe98
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/blastxml_to_gapped_gff3 commit 8f38145c94ecb1e23c3ff6f0243213dc49d2287e
 iuc parents: diff
changeset | 134 | 
| 
bd47051afe98
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/blastxml_to_gapped_gff3 commit 8f38145c94ecb1e23c3ff6f0243213dc49d2287e
 iuc parents: diff
changeset | 135 Q:ACTGACTGACTG | 
| 
bd47051afe98
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/blastxml_to_gapped_gff3 commit 8f38145c94ecb1e23c3ff6f0243213dc49d2287e
 iuc parents: diff
changeset | 136 S:ACTGAC---CTG | 
| 
bd47051afe98
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/blastxml_to_gapped_gff3 commit 8f38145c94ecb1e23c3ff6f0243213dc49d2287e
 iuc parents: diff
changeset | 137 | 
| 
bd47051afe98
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/blastxml_to_gapped_gff3 commit 8f38145c94ecb1e23c3ff6f0243213dc49d2287e
 iuc parents: diff
changeset | 138 which greatly simplifies the process of identifying the correct location | 
| 
bd47051afe98
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/blastxml_to_gapped_gff3 commit 8f38145c94ecb1e23c3ff6f0243213dc49d2287e
 iuc parents: diff
changeset | 139 for a match_part | 
| 
bd47051afe98
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/blastxml_to_gapped_gff3 commit 8f38145c94ecb1e23c3ff6f0243213dc49d2287e
 iuc parents: diff
changeset | 140 """ | 
| 
bd47051afe98
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/blastxml_to_gapped_gff3 commit 8f38145c94ecb1e23c3ff6f0243213dc49d2287e
 iuc parents: diff
changeset | 141 prev = 0 | 
| 
bd47051afe98
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/blastxml_to_gapped_gff3 commit 8f38145c94ecb1e23c3ff6f0243213dc49d2287e
 iuc parents: diff
changeset | 142 fq = '' | 
| 
bd47051afe98
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/blastxml_to_gapped_gff3 commit 8f38145c94ecb1e23c3ff6f0243213dc49d2287e
 iuc parents: diff
changeset | 143 fm = '' | 
| 
bd47051afe98
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/blastxml_to_gapped_gff3 commit 8f38145c94ecb1e23c3ff6f0243213dc49d2287e
 iuc parents: diff
changeset | 144 fs = '' | 
| 
bd47051afe98
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/blastxml_to_gapped_gff3 commit 8f38145c94ecb1e23c3ff6f0243213dc49d2287e
 iuc parents: diff
changeset | 145 for position in re.finditer('-', query): | 
| 
bd47051afe98
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/blastxml_to_gapped_gff3 commit 8f38145c94ecb1e23c3ff6f0243213dc49d2287e
 iuc parents: diff
changeset | 146 fq += query[prev:position.start()] | 
| 
bd47051afe98
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/blastxml_to_gapped_gff3 commit 8f38145c94ecb1e23c3ff6f0243213dc49d2287e
 iuc parents: diff
changeset | 147 fm += match[prev:position.start()] | 
| 
bd47051afe98
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/blastxml_to_gapped_gff3 commit 8f38145c94ecb1e23c3ff6f0243213dc49d2287e
 iuc parents: diff
changeset | 148 fs += subject[prev:position.start()] | 
| 
bd47051afe98
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/blastxml_to_gapped_gff3 commit 8f38145c94ecb1e23c3ff6f0243213dc49d2287e
 iuc parents: diff
changeset | 149 prev = position.start() + 1 | 
| 
bd47051afe98
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/blastxml_to_gapped_gff3 commit 8f38145c94ecb1e23c3ff6f0243213dc49d2287e
 iuc parents: diff
changeset | 150 fq += query[prev:] | 
| 
bd47051afe98
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/blastxml_to_gapped_gff3 commit 8f38145c94ecb1e23c3ff6f0243213dc49d2287e
 iuc parents: diff
changeset | 151 fm += match[prev:] | 
| 
bd47051afe98
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/blastxml_to_gapped_gff3 commit 8f38145c94ecb1e23c3ff6f0243213dc49d2287e
 iuc parents: diff
changeset | 152 fs += subject[prev:] | 
| 
bd47051afe98
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/blastxml_to_gapped_gff3 commit 8f38145c94ecb1e23c3ff6f0243213dc49d2287e
 iuc parents: diff
changeset | 153 | 
| 
bd47051afe98
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/blastxml_to_gapped_gff3 commit 8f38145c94ecb1e23c3ff6f0243213dc49d2287e
 iuc parents: diff
changeset | 154 return (fq, fm, fs) | 
| 
bd47051afe98
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/blastxml_to_gapped_gff3 commit 8f38145c94ecb1e23c3ff6f0243213dc49d2287e
 iuc parents: diff
changeset | 155 | 
| 
bd47051afe98
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/blastxml_to_gapped_gff3 commit 8f38145c94ecb1e23c3ff6f0243213dc49d2287e
 iuc parents: diff
changeset | 156 | 
| 
bd47051afe98
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/blastxml_to_gapped_gff3 commit 8f38145c94ecb1e23c3ff6f0243213dc49d2287e
 iuc parents: diff
changeset | 157 def generate_parts(query, match, subject, ignore_under=3): | 
| 
bd47051afe98
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/blastxml_to_gapped_gff3 commit 8f38145c94ecb1e23c3ff6f0243213dc49d2287e
 iuc parents: diff
changeset | 158 region_q = [] | 
| 
bd47051afe98
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/blastxml_to_gapped_gff3 commit 8f38145c94ecb1e23c3ff6f0243213dc49d2287e
 iuc parents: diff
changeset | 159 region_m = [] | 
| 
bd47051afe98
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/blastxml_to_gapped_gff3 commit 8f38145c94ecb1e23c3ff6f0243213dc49d2287e
 iuc parents: diff
changeset | 160 region_s = [] | 
| 
bd47051afe98
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/blastxml_to_gapped_gff3 commit 8f38145c94ecb1e23c3ff6f0243213dc49d2287e
 iuc parents: diff
changeset | 161 | 
| 
bd47051afe98
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/blastxml_to_gapped_gff3 commit 8f38145c94ecb1e23c3ff6f0243213dc49d2287e
 iuc parents: diff
changeset | 162 (query, match, subject) = __remove_query_gaps(query, match, subject) | 
| 
bd47051afe98
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/blastxml_to_gapped_gff3 commit 8f38145c94ecb1e23c3ff6f0243213dc49d2287e
 iuc parents: diff
changeset | 163 | 
| 
bd47051afe98
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/blastxml_to_gapped_gff3 commit 8f38145c94ecb1e23c3ff6f0243213dc49d2287e
 iuc parents: diff
changeset | 164 region_start = -1 | 
| 
bd47051afe98
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/blastxml_to_gapped_gff3 commit 8f38145c94ecb1e23c3ff6f0243213dc49d2287e
 iuc parents: diff
changeset | 165 region_end = -1 | 
| 
bd47051afe98
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/blastxml_to_gapped_gff3 commit 8f38145c94ecb1e23c3ff6f0243213dc49d2287e
 iuc parents: diff
changeset | 166 mismatch_count = 0 | 
| 
bd47051afe98
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/blastxml_to_gapped_gff3 commit 8f38145c94ecb1e23c3ff6f0243213dc49d2287e
 iuc parents: diff
changeset | 167 for i, (q, m, s) in enumerate(zip(query, match, subject)): | 
| 
bd47051afe98
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/blastxml_to_gapped_gff3 commit 8f38145c94ecb1e23c3ff6f0243213dc49d2287e
 iuc parents: diff
changeset | 168 | 
| 
bd47051afe98
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/blastxml_to_gapped_gff3 commit 8f38145c94ecb1e23c3ff6f0243213dc49d2287e
 iuc parents: diff
changeset | 169 # If we have a match | 
| 
bd47051afe98
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/blastxml_to_gapped_gff3 commit 8f38145c94ecb1e23c3ff6f0243213dc49d2287e
 iuc parents: diff
changeset | 170 if m != ' ' or m == '+': | 
| 
bd47051afe98
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/blastxml_to_gapped_gff3 commit 8f38145c94ecb1e23c3ff6f0243213dc49d2287e
 iuc parents: diff
changeset | 171 if region_start == -1: | 
| 
bd47051afe98
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/blastxml_to_gapped_gff3 commit 8f38145c94ecb1e23c3ff6f0243213dc49d2287e
 iuc parents: diff
changeset | 172 region_start = i | 
| 
bd47051afe98
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/blastxml_to_gapped_gff3 commit 8f38145c94ecb1e23c3ff6f0243213dc49d2287e
 iuc parents: diff
changeset | 173 # It's a new region, we need to reset or it's pre-seeded with | 
| 
bd47051afe98
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/blastxml_to_gapped_gff3 commit 8f38145c94ecb1e23c3ff6f0243213dc49d2287e
 iuc parents: diff
changeset | 174 # spaces | 
| 
bd47051afe98
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/blastxml_to_gapped_gff3 commit 8f38145c94ecb1e23c3ff6f0243213dc49d2287e
 iuc parents: diff
changeset | 175 region_q = [] | 
| 
bd47051afe98
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/blastxml_to_gapped_gff3 commit 8f38145c94ecb1e23c3ff6f0243213dc49d2287e
 iuc parents: diff
changeset | 176 region_m = [] | 
| 
bd47051afe98
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/blastxml_to_gapped_gff3 commit 8f38145c94ecb1e23c3ff6f0243213dc49d2287e
 iuc parents: diff
changeset | 177 region_s = [] | 
| 
bd47051afe98
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/blastxml_to_gapped_gff3 commit 8f38145c94ecb1e23c3ff6f0243213dc49d2287e
 iuc parents: diff
changeset | 178 region_end = i | 
| 
bd47051afe98
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/blastxml_to_gapped_gff3 commit 8f38145c94ecb1e23c3ff6f0243213dc49d2287e
 iuc parents: diff
changeset | 179 mismatch_count = 0 | 
| 
bd47051afe98
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/blastxml_to_gapped_gff3 commit 8f38145c94ecb1e23c3ff6f0243213dc49d2287e
 iuc parents: diff
changeset | 180 else: | 
| 
bd47051afe98
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/blastxml_to_gapped_gff3 commit 8f38145c94ecb1e23c3ff6f0243213dc49d2287e
 iuc parents: diff
changeset | 181 mismatch_count += 1 | 
| 
bd47051afe98
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/blastxml_to_gapped_gff3 commit 8f38145c94ecb1e23c3ff6f0243213dc49d2287e
 iuc parents: diff
changeset | 182 | 
| 
bd47051afe98
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/blastxml_to_gapped_gff3 commit 8f38145c94ecb1e23c3ff6f0243213dc49d2287e
 iuc parents: diff
changeset | 183 region_q.append(q) | 
| 
bd47051afe98
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/blastxml_to_gapped_gff3 commit 8f38145c94ecb1e23c3ff6f0243213dc49d2287e
 iuc parents: diff
changeset | 184 region_m.append(m) | 
| 
bd47051afe98
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/blastxml_to_gapped_gff3 commit 8f38145c94ecb1e23c3ff6f0243213dc49d2287e
 iuc parents: diff
changeset | 185 region_s.append(s) | 
| 
bd47051afe98
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/blastxml_to_gapped_gff3 commit 8f38145c94ecb1e23c3ff6f0243213dc49d2287e
 iuc parents: diff
changeset | 186 | 
| 
bd47051afe98
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/blastxml_to_gapped_gff3 commit 8f38145c94ecb1e23c3ff6f0243213dc49d2287e
 iuc parents: diff
changeset | 187 if mismatch_count >= ignore_under and region_start != -1 and region_end != -1: | 
| 
bd47051afe98
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/blastxml_to_gapped_gff3 commit 8f38145c94ecb1e23c3ff6f0243213dc49d2287e
 iuc parents: diff
changeset | 188 region_q = region_q[0:-ignore_under] | 
| 
bd47051afe98
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/blastxml_to_gapped_gff3 commit 8f38145c94ecb1e23c3ff6f0243213dc49d2287e
 iuc parents: diff
changeset | 189 region_m = region_m[0:-ignore_under] | 
| 
bd47051afe98
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/blastxml_to_gapped_gff3 commit 8f38145c94ecb1e23c3ff6f0243213dc49d2287e
 iuc parents: diff
changeset | 190 region_s = region_s[0:-ignore_under] | 
| 
bd47051afe98
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/blastxml_to_gapped_gff3 commit 8f38145c94ecb1e23c3ff6f0243213dc49d2287e
 iuc parents: diff
changeset | 191 yield region_start, region_end + 1, \ | 
| 
bd47051afe98
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/blastxml_to_gapped_gff3 commit 8f38145c94ecb1e23c3ff6f0243213dc49d2287e
 iuc parents: diff
changeset | 192 cigar_from_string(region_q, region_m, region_s, strict_m=True) | 
| 
bd47051afe98
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/blastxml_to_gapped_gff3 commit 8f38145c94ecb1e23c3ff6f0243213dc49d2287e
 iuc parents: diff
changeset | 193 region_q = [] | 
| 
bd47051afe98
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/blastxml_to_gapped_gff3 commit 8f38145c94ecb1e23c3ff6f0243213dc49d2287e
 iuc parents: diff
changeset | 194 region_m = [] | 
| 
bd47051afe98
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/blastxml_to_gapped_gff3 commit 8f38145c94ecb1e23c3ff6f0243213dc49d2287e
 iuc parents: diff
changeset | 195 region_s = [] | 
| 
bd47051afe98
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/blastxml_to_gapped_gff3 commit 8f38145c94ecb1e23c3ff6f0243213dc49d2287e
 iuc parents: diff
changeset | 196 | 
| 
bd47051afe98
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/blastxml_to_gapped_gff3 commit 8f38145c94ecb1e23c3ff6f0243213dc49d2287e
 iuc parents: diff
changeset | 197 region_start = -1 | 
| 
bd47051afe98
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/blastxml_to_gapped_gff3 commit 8f38145c94ecb1e23c3ff6f0243213dc49d2287e
 iuc parents: diff
changeset | 198 region_end = -1 | 
| 
bd47051afe98
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/blastxml_to_gapped_gff3 commit 8f38145c94ecb1e23c3ff6f0243213dc49d2287e
 iuc parents: diff
changeset | 199 mismatch_count = 0 | 
| 
bd47051afe98
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/blastxml_to_gapped_gff3 commit 8f38145c94ecb1e23c3ff6f0243213dc49d2287e
 iuc parents: diff
changeset | 200 | 
| 
bd47051afe98
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/blastxml_to_gapped_gff3 commit 8f38145c94ecb1e23c3ff6f0243213dc49d2287e
 iuc parents: diff
changeset | 201 yield region_start, region_end + 1, \ | 
| 
bd47051afe98
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/blastxml_to_gapped_gff3 commit 8f38145c94ecb1e23c3ff6f0243213dc49d2287e
 iuc parents: diff
changeset | 202 cigar_from_string(region_q, region_m, region_s, strict_m=True) | 
| 
bd47051afe98
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/blastxml_to_gapped_gff3 commit 8f38145c94ecb1e23c3ff6f0243213dc49d2287e
 iuc parents: diff
changeset | 203 | 
| 
bd47051afe98
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/blastxml_to_gapped_gff3 commit 8f38145c94ecb1e23c3ff6f0243213dc49d2287e
 iuc parents: diff
changeset | 204 | 
| 
bd47051afe98
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/blastxml_to_gapped_gff3 commit 8f38145c94ecb1e23c3ff6f0243213dc49d2287e
 iuc parents: diff
changeset | 205 def _qms_to_matches(query, match, subject, strict_m=True): | 
| 
bd47051afe98
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/blastxml_to_gapped_gff3 commit 8f38145c94ecb1e23c3ff6f0243213dc49d2287e
 iuc parents: diff
changeset | 206 matchline = [] | 
| 
bd47051afe98
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/blastxml_to_gapped_gff3 commit 8f38145c94ecb1e23c3ff6f0243213dc49d2287e
 iuc parents: diff
changeset | 207 | 
| 
bd47051afe98
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/blastxml_to_gapped_gff3 commit 8f38145c94ecb1e23c3ff6f0243213dc49d2287e
 iuc parents: diff
changeset | 208 for (q, m, s) in zip(query, match, subject): | 
| 
bd47051afe98
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/blastxml_to_gapped_gff3 commit 8f38145c94ecb1e23c3ff6f0243213dc49d2287e
 iuc parents: diff
changeset | 209 ret = '' | 
| 
bd47051afe98
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/blastxml_to_gapped_gff3 commit 8f38145c94ecb1e23c3ff6f0243213dc49d2287e
 iuc parents: diff
changeset | 210 | 
| 
bd47051afe98
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/blastxml_to_gapped_gff3 commit 8f38145c94ecb1e23c3ff6f0243213dc49d2287e
 iuc parents: diff
changeset | 211 if m != ' ' or m == '+': | 
| 
bd47051afe98
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/blastxml_to_gapped_gff3 commit 8f38145c94ecb1e23c3ff6f0243213dc49d2287e
 iuc parents: diff
changeset | 212 ret = '=' | 
| 
bd47051afe98
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/blastxml_to_gapped_gff3 commit 8f38145c94ecb1e23c3ff6f0243213dc49d2287e
 iuc parents: diff
changeset | 213 elif m == ' ': | 
| 
bd47051afe98
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/blastxml_to_gapped_gff3 commit 8f38145c94ecb1e23c3ff6f0243213dc49d2287e
 iuc parents: diff
changeset | 214 if q == '-': | 
| 
bd47051afe98
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/blastxml_to_gapped_gff3 commit 8f38145c94ecb1e23c3ff6f0243213dc49d2287e
 iuc parents: diff
changeset | 215 ret = 'D' | 
| 
bd47051afe98
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/blastxml_to_gapped_gff3 commit 8f38145c94ecb1e23c3ff6f0243213dc49d2287e
 iuc parents: diff
changeset | 216 elif s == '-': | 
| 
bd47051afe98
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/blastxml_to_gapped_gff3 commit 8f38145c94ecb1e23c3ff6f0243213dc49d2287e
 iuc parents: diff
changeset | 217 ret = 'I' | 
| 
bd47051afe98
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/blastxml_to_gapped_gff3 commit 8f38145c94ecb1e23c3ff6f0243213dc49d2287e
 iuc parents: diff
changeset | 218 else: | 
| 
bd47051afe98
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/blastxml_to_gapped_gff3 commit 8f38145c94ecb1e23c3ff6f0243213dc49d2287e
 iuc parents: diff
changeset | 219 ret = 'X' | 
| 
bd47051afe98
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/blastxml_to_gapped_gff3 commit 8f38145c94ecb1e23c3ff6f0243213dc49d2287e
 iuc parents: diff
changeset | 220 else: | 
| 
bd47051afe98
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/blastxml_to_gapped_gff3 commit 8f38145c94ecb1e23c3ff6f0243213dc49d2287e
 iuc parents: diff
changeset | 221 log.warn("Bad data: \n\t%s\n\t%s\n\t%s\n" % (query, match, subject)) | 
| 
bd47051afe98
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/blastxml_to_gapped_gff3 commit 8f38145c94ecb1e23c3ff6f0243213dc49d2287e
 iuc parents: diff
changeset | 222 | 
| 
bd47051afe98
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/blastxml_to_gapped_gff3 commit 8f38145c94ecb1e23c3ff6f0243213dc49d2287e
 iuc parents: diff
changeset | 223 if strict_m: | 
| 
bd47051afe98
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/blastxml_to_gapped_gff3 commit 8f38145c94ecb1e23c3ff6f0243213dc49d2287e
 iuc parents: diff
changeset | 224 if ret == '=' or ret == 'X': | 
| 
bd47051afe98
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/blastxml_to_gapped_gff3 commit 8f38145c94ecb1e23c3ff6f0243213dc49d2287e
 iuc parents: diff
changeset | 225 ret = 'M' | 
| 
bd47051afe98
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/blastxml_to_gapped_gff3 commit 8f38145c94ecb1e23c3ff6f0243213dc49d2287e
 iuc parents: diff
changeset | 226 | 
| 
bd47051afe98
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/blastxml_to_gapped_gff3 commit 8f38145c94ecb1e23c3ff6f0243213dc49d2287e
 iuc parents: diff
changeset | 227 matchline.append(ret) | 
| 
bd47051afe98
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/blastxml_to_gapped_gff3 commit 8f38145c94ecb1e23c3ff6f0243213dc49d2287e
 iuc parents: diff
changeset | 228 return matchline | 
| 
bd47051afe98
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/blastxml_to_gapped_gff3 commit 8f38145c94ecb1e23c3ff6f0243213dc49d2287e
 iuc parents: diff
changeset | 229 | 
| 
bd47051afe98
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/blastxml_to_gapped_gff3 commit 8f38145c94ecb1e23c3ff6f0243213dc49d2287e
 iuc parents: diff
changeset | 230 | 
| 
bd47051afe98
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/blastxml_to_gapped_gff3 commit 8f38145c94ecb1e23c3ff6f0243213dc49d2287e
 iuc parents: diff
changeset | 231 def _matchline_to_cigar(matchline): | 
| 
bd47051afe98
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/blastxml_to_gapped_gff3 commit 8f38145c94ecb1e23c3ff6f0243213dc49d2287e
 iuc parents: diff
changeset | 232 cigar_line = [] | 
| 
bd47051afe98
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/blastxml_to_gapped_gff3 commit 8f38145c94ecb1e23c3ff6f0243213dc49d2287e
 iuc parents: diff
changeset | 233 last_char = matchline[0] | 
| 
bd47051afe98
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/blastxml_to_gapped_gff3 commit 8f38145c94ecb1e23c3ff6f0243213dc49d2287e
 iuc parents: diff
changeset | 234 count = 0 | 
| 
bd47051afe98
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/blastxml_to_gapped_gff3 commit 8f38145c94ecb1e23c3ff6f0243213dc49d2287e
 iuc parents: diff
changeset | 235 for char in matchline: | 
| 
bd47051afe98
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/blastxml_to_gapped_gff3 commit 8f38145c94ecb1e23c3ff6f0243213dc49d2287e
 iuc parents: diff
changeset | 236 if char == last_char: | 
| 
bd47051afe98
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/blastxml_to_gapped_gff3 commit 8f38145c94ecb1e23c3ff6f0243213dc49d2287e
 iuc parents: diff
changeset | 237 count += 1 | 
| 
bd47051afe98
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/blastxml_to_gapped_gff3 commit 8f38145c94ecb1e23c3ff6f0243213dc49d2287e
 iuc parents: diff
changeset | 238 else: | 
| 
bd47051afe98
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/blastxml_to_gapped_gff3 commit 8f38145c94ecb1e23c3ff6f0243213dc49d2287e
 iuc parents: diff
changeset | 239 cigar_line.append("%s%s" % (last_char, count)) | 
| 
bd47051afe98
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/blastxml_to_gapped_gff3 commit 8f38145c94ecb1e23c3ff6f0243213dc49d2287e
 iuc parents: diff
changeset | 240 count = 1 | 
| 
bd47051afe98
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/blastxml_to_gapped_gff3 commit 8f38145c94ecb1e23c3ff6f0243213dc49d2287e
 iuc parents: diff
changeset | 241 last_char = char | 
| 
bd47051afe98
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/blastxml_to_gapped_gff3 commit 8f38145c94ecb1e23c3ff6f0243213dc49d2287e
 iuc parents: diff
changeset | 242 cigar_line.append("%s%s" % (last_char, count)) | 
| 
bd47051afe98
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/blastxml_to_gapped_gff3 commit 8f38145c94ecb1e23c3ff6f0243213dc49d2287e
 iuc parents: diff
changeset | 243 return ' '.join(cigar_line) | 
| 
bd47051afe98
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/blastxml_to_gapped_gff3 commit 8f38145c94ecb1e23c3ff6f0243213dc49d2287e
 iuc parents: diff
changeset | 244 | 
| 
bd47051afe98
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/blastxml_to_gapped_gff3 commit 8f38145c94ecb1e23c3ff6f0243213dc49d2287e
 iuc parents: diff
changeset | 245 | 
| 
bd47051afe98
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/blastxml_to_gapped_gff3 commit 8f38145c94ecb1e23c3ff6f0243213dc49d2287e
 iuc parents: diff
changeset | 246 def cigar_from_string(query, match, subject, strict_m=True): | 
| 
bd47051afe98
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/blastxml_to_gapped_gff3 commit 8f38145c94ecb1e23c3ff6f0243213dc49d2287e
 iuc parents: diff
changeset | 247 matchline = _qms_to_matches(query, match, subject, strict_m=strict_m) | 
| 
bd47051afe98
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/blastxml_to_gapped_gff3 commit 8f38145c94ecb1e23c3ff6f0243213dc49d2287e
 iuc parents: diff
changeset | 248 if len(matchline) > 0: | 
| 
bd47051afe98
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/blastxml_to_gapped_gff3 commit 8f38145c94ecb1e23c3ff6f0243213dc49d2287e
 iuc parents: diff
changeset | 249 return _matchline_to_cigar(matchline) | 
| 
bd47051afe98
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/blastxml_to_gapped_gff3 commit 8f38145c94ecb1e23c3ff6f0243213dc49d2287e
 iuc parents: diff
changeset | 250 else: | 
| 
bd47051afe98
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/blastxml_to_gapped_gff3 commit 8f38145c94ecb1e23c3ff6f0243213dc49d2287e
 iuc parents: diff
changeset | 251 return "" | 
| 
bd47051afe98
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/blastxml_to_gapped_gff3 commit 8f38145c94ecb1e23c3ff6f0243213dc49d2287e
 iuc parents: diff
changeset | 252 | 
| 
bd47051afe98
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/blastxml_to_gapped_gff3 commit 8f38145c94ecb1e23c3ff6f0243213dc49d2287e
 iuc parents: diff
changeset | 253 | 
| 
bd47051afe98
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/blastxml_to_gapped_gff3 commit 8f38145c94ecb1e23c3ff6f0243213dc49d2287e
 iuc parents: diff
changeset | 254 if __name__ == '__main__': | 
| 
bd47051afe98
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/blastxml_to_gapped_gff3 commit 8f38145c94ecb1e23c3ff6f0243213dc49d2287e
 iuc parents: diff
changeset | 255 parser = argparse.ArgumentParser(description='Convert Blast XML to gapped GFF3', epilog='') | 
| 
bd47051afe98
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/blastxml_to_gapped_gff3 commit 8f38145c94ecb1e23c3ff6f0243213dc49d2287e
 iuc parents: diff
changeset | 256 parser.add_argument('blastxml', type=open, help='Blast XML Output') | 
| 
bd47051afe98
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/blastxml_to_gapped_gff3 commit 8f38145c94ecb1e23c3ff6f0243213dc49d2287e
 iuc parents: diff
changeset | 257 parser.add_argument('--min_gap', type=int, help='Maximum gap size before generating a new match_part', default=3) | 
| 
bd47051afe98
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/blastxml_to_gapped_gff3 commit 8f38145c94ecb1e23c3ff6f0243213dc49d2287e
 iuc parents: diff
changeset | 258 parser.add_argument('--trim', action='store_true', help='Trim blast hits to be only as long as the parent feature') | 
| 
bd47051afe98
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/blastxml_to_gapped_gff3 commit 8f38145c94ecb1e23c3ff6f0243213dc49d2287e
 iuc parents: diff
changeset | 259 parser.add_argument('--trim_end', action='store_true', help='Cut blast results off at end of gene') | 
| 
bd47051afe98
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/blastxml_to_gapped_gff3 commit 8f38145c94ecb1e23c3ff6f0243213dc49d2287e
 iuc parents: diff
changeset | 260 args = parser.parse_args() | 
| 
bd47051afe98
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/blastxml_to_gapped_gff3 commit 8f38145c94ecb1e23c3ff6f0243213dc49d2287e
 iuc parents: diff
changeset | 261 | 
| 
bd47051afe98
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/blastxml_to_gapped_gff3 commit 8f38145c94ecb1e23c3ff6f0243213dc49d2287e
 iuc parents: diff
changeset | 262 result = blastxml2gff3(**vars(args)) | 
| 
bd47051afe98
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/blastxml_to_gapped_gff3 commit 8f38145c94ecb1e23c3ff6f0243213dc49d2287e
 iuc parents: diff
changeset | 263 GFF.write(result, sys.stdout) | 
