Mercurial > repos > eric-rasche > apollo

annotate gff3_cleaner.py @ 2:c8e16c8eff98 draft

Find changesets by keywords (author, files, the commit message), revision number or hash, or revset expression.
planemo upload for repository https://github.com/TAMU-CPT/galaxy-webapollo commit 6944305517dcccad3e6b37f8aba0b43ccb004540
author eric-rasche
date Tue, 03 May 2016 13:48:11 -0400
parents 6002cc0df04e
children
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
rev   line source
0
6002cc0df04e planemo upload for repository https://github.com/TAMU-CPT/galaxy-webapollo commit 4e5a5af7689f1713c34a6ad9a9594c205e762fdd
eric-rasche
parents:
diff changeset
1 #!/usr/bin/env python
6002cc0df04e planemo upload for repository https://github.com/TAMU-CPT/galaxy-webapollo commit 4e5a5af7689f1713c34a6ad9a9594c205e762fdd
eric-rasche
parents:
diff changeset
2 import sys
6002cc0df04e planemo upload for repository https://github.com/TAMU-CPT/galaxy-webapollo commit 4e5a5af7689f1713c34a6ad9a9594c205e762fdd
eric-rasche
parents:
diff changeset
3 import argparse
6002cc0df04e planemo upload for repository https://github.com/TAMU-CPT/galaxy-webapollo commit 4e5a5af7689f1713c34a6ad9a9594c205e762fdd
eric-rasche
parents:
diff changeset
4 from gff3 import feature_lambda, feature_test_type
6002cc0df04e planemo upload for repository https://github.com/TAMU-CPT/galaxy-webapollo commit 4e5a5af7689f1713c34a6ad9a9594c205e762fdd
eric-rasche
parents:
diff changeset
5 from BCBio import GFF
6002cc0df04e planemo upload for repository https://github.com/TAMU-CPT/galaxy-webapollo commit 4e5a5af7689f1713c34a6ad9a9594c205e762fdd
eric-rasche
parents:
diff changeset
6 import logging
6002cc0df04e planemo upload for repository https://github.com/TAMU-CPT/galaxy-webapollo commit 4e5a5af7689f1713c34a6ad9a9594c205e762fdd
eric-rasche
parents:
diff changeset
7 logging.basicConfig(level=logging.WARN)
6002cc0df04e planemo upload for repository https://github.com/TAMU-CPT/galaxy-webapollo commit 4e5a5af7689f1713c34a6ad9a9594c205e762fdd
eric-rasche
parents:
diff changeset
8 log = logging.getLogger(name='pav')
6002cc0df04e planemo upload for repository https://github.com/TAMU-CPT/galaxy-webapollo commit 4e5a5af7689f1713c34a6ad9a9594c205e762fdd
eric-rasche
parents:
diff changeset
9
6002cc0df04e planemo upload for repository https://github.com/TAMU-CPT/galaxy-webapollo commit 4e5a5af7689f1713c34a6ad9a9594c205e762fdd
eric-rasche
parents:
diff changeset
10 def coding_genes(feature_list):
6002cc0df04e planemo upload for repository https://github.com/TAMU-CPT/galaxy-webapollo commit 4e5a5af7689f1713c34a6ad9a9594c205e762fdd
eric-rasche
parents:
diff changeset
11 for x in feature_lambda(feature_list, feature_test_type, {'type': 'gene'}, subfeatures=True):
6002cc0df04e planemo upload for repository https://github.com/TAMU-CPT/galaxy-webapollo commit 4e5a5af7689f1713c34a6ad9a9594c205e762fdd
eric-rasche
parents:
diff changeset
12 if len(list(feature_lambda(x.sub_features, feature_test_type, {'type': 'CDS'}, subfeatures=False))) > 0:
6002cc0df04e planemo upload for repository https://github.com/TAMU-CPT/galaxy-webapollo commit 4e5a5af7689f1713c34a6ad9a9594c205e762fdd
eric-rasche
parents:
diff changeset
13 yield x
6002cc0df04e planemo upload for repository https://github.com/TAMU-CPT/galaxy-webapollo commit 4e5a5af7689f1713c34a6ad9a9594c205e762fdd
eric-rasche
parents:
diff changeset
14
6002cc0df04e planemo upload for repository https://github.com/TAMU-CPT/galaxy-webapollo commit 4e5a5af7689f1713c34a6ad9a9594c205e762fdd
eric-rasche
parents:
diff changeset
15
6002cc0df04e planemo upload for repository https://github.com/TAMU-CPT/galaxy-webapollo commit 4e5a5af7689f1713c34a6ad9a9594c205e762fdd
eric-rasche
parents:
diff changeset
16 def genes(feature_list, feature_type='gene'):
6002cc0df04e planemo upload for repository https://github.com/TAMU-CPT/galaxy-webapollo commit 4e5a5af7689f1713c34a6ad9a9594c205e762fdd
eric-rasche
parents:
diff changeset
17 for x in feature_lambda(feature_list, feature_test_type,
6002cc0df04e planemo upload for repository https://github.com/TAMU-CPT/galaxy-webapollo commit 4e5a5af7689f1713c34a6ad9a9594c205e762fdd
eric-rasche
parents:
diff changeset
18 {'type': feature_type},
6002cc0df04e planemo upload for repository https://github.com/TAMU-CPT/galaxy-webapollo commit 4e5a5af7689f1713c34a6ad9a9594c205e762fdd
eric-rasche
parents:
diff changeset
19 subfeatures=True):
6002cc0df04e planemo upload for repository https://github.com/TAMU-CPT/galaxy-webapollo commit 4e5a5af7689f1713c34a6ad9a9594c205e762fdd
eric-rasche
parents:
diff changeset
20 yield x
6002cc0df04e planemo upload for repository https://github.com/TAMU-CPT/galaxy-webapollo commit 4e5a5af7689f1713c34a6ad9a9594c205e762fdd
eric-rasche
parents:
diff changeset
21
6002cc0df04e planemo upload for repository https://github.com/TAMU-CPT/galaxy-webapollo commit 4e5a5af7689f1713c34a6ad9a9594c205e762fdd
eric-rasche
parents:
diff changeset
22
6002cc0df04e planemo upload for repository https://github.com/TAMU-CPT/galaxy-webapollo commit 4e5a5af7689f1713c34a6ad9a9594c205e762fdd
eric-rasche
parents:
diff changeset
23 def fix_apollo_issues(annotations, user_email):
6002cc0df04e planemo upload for repository https://github.com/TAMU-CPT/galaxy-webapollo commit 4e5a5af7689f1713c34a6ad9a9594c205e762fdd
eric-rasche
parents:
diff changeset
24 for rec in GFF.parse(annotations):
6002cc0df04e planemo upload for repository https://github.com/TAMU-CPT/galaxy-webapollo commit 4e5a5af7689f1713c34a6ad9a9594c205e762fdd
eric-rasche
parents:
diff changeset
25 for feat in rec.features:
6002cc0df04e planemo upload for repository https://github.com/TAMU-CPT/galaxy-webapollo commit 4e5a5af7689f1713c34a6ad9a9594c205e762fdd
eric-rasche
parents:
diff changeset
26 if feat.type != 'gene':
6002cc0df04e planemo upload for repository https://github.com/TAMU-CPT/galaxy-webapollo commit 4e5a5af7689f1713c34a6ad9a9594c205e762fdd
eric-rasche
parents:
diff changeset
27 continue
6002cc0df04e planemo upload for repository https://github.com/TAMU-CPT/galaxy-webapollo commit 4e5a5af7689f1713c34a6ad9a9594c205e762fdd
eric-rasche
parents:
diff changeset
28
6002cc0df04e planemo upload for repository https://github.com/TAMU-CPT/galaxy-webapollo commit 4e5a5af7689f1713c34a6ad9a9594c205e762fdd
eric-rasche
parents:
diff changeset
29 for sf in feat.sub_features:
6002cc0df04e planemo upload for repository https://github.com/TAMU-CPT/galaxy-webapollo commit 4e5a5af7689f1713c34a6ad9a9594c205e762fdd
eric-rasche
parents:
diff changeset
30 if sf.type != 'mRNA':
6002cc0df04e planemo upload for repository https://github.com/TAMU-CPT/galaxy-webapollo commit 4e5a5af7689f1713c34a6ad9a9594c205e762fdd
eric-rasche
parents:
diff changeset
31 continue
6002cc0df04e planemo upload for repository https://github.com/TAMU-CPT/galaxy-webapollo commit 4e5a5af7689f1713c34a6ad9a9594c205e762fdd
eric-rasche
parents:
diff changeset
32
6002cc0df04e planemo upload for repository https://github.com/TAMU-CPT/galaxy-webapollo commit 4e5a5af7689f1713c34a6ad9a9594c205e762fdd
eric-rasche
parents:
diff changeset
33 for ssf in sf.sub_features:
6002cc0df04e planemo upload for repository https://github.com/TAMU-CPT/galaxy-webapollo commit 4e5a5af7689f1713c34a6ad9a9594c205e762fdd
eric-rasche
parents:
diff changeset
34 if ssf.type != 'exon':
6002cc0df04e planemo upload for repository https://github.com/TAMU-CPT/galaxy-webapollo commit 4e5a5af7689f1713c34a6ad9a9594c205e762fdd
eric-rasche
parents:
diff changeset
35 continue
6002cc0df04e planemo upload for repository https://github.com/TAMU-CPT/galaxy-webapollo commit 4e5a5af7689f1713c34a6ad9a9594c205e762fdd
eric-rasche
parents:
diff changeset
36
6002cc0df04e planemo upload for repository https://github.com/TAMU-CPT/galaxy-webapollo commit 4e5a5af7689f1713c34a6ad9a9594c205e762fdd
eric-rasche
parents:
diff changeset
37 if len(ssf) > 10:
6002cc0df04e planemo upload for repository https://github.com/TAMU-CPT/galaxy-webapollo commit 4e5a5af7689f1713c34a6ad9a9594c205e762fdd
eric-rasche
parents:
diff changeset
38 continue
6002cc0df04e planemo upload for repository https://github.com/TAMU-CPT/galaxy-webapollo commit 4e5a5af7689f1713c34a6ad9a9594c205e762fdd
eric-rasche
parents:
diff changeset
39
6002cc0df04e planemo upload for repository https://github.com/TAMU-CPT/galaxy-webapollo commit 4e5a5af7689f1713c34a6ad9a9594c205e762fdd
eric-rasche
parents:
diff changeset
40 ssf.type = 'Shine_Dalgarno_sequence'
6002cc0df04e planemo upload for repository https://github.com/TAMU-CPT/galaxy-webapollo commit 4e5a5af7689f1713c34a6ad9a9594c205e762fdd
eric-rasche
parents:
diff changeset
41
6002cc0df04e planemo upload for repository https://github.com/TAMU-CPT/galaxy-webapollo commit 4e5a5af7689f1713c34a6ad9a9594c205e762fdd
eric-rasche
parents:
diff changeset
42 sf.sub_features = [x for x in sf.sub_features if x.type not in
6002cc0df04e planemo upload for repository https://github.com/TAMU-CPT/galaxy-webapollo commit 4e5a5af7689f1713c34a6ad9a9594c205e762fdd
eric-rasche
parents:
diff changeset
43 ('non_canonical_five_prime_splice_site',
6002cc0df04e planemo upload for repository https://github.com/TAMU-CPT/galaxy-webapollo commit 4e5a5af7689f1713c34a6ad9a9594c205e762fdd
eric-rasche
parents:
diff changeset
44 'non_canonical_three_prime_splice_site')]
6002cc0df04e planemo upload for repository https://github.com/TAMU-CPT/galaxy-webapollo commit 4e5a5af7689f1713c34a6ad9a9594c205e762fdd
eric-rasche
parents:
diff changeset
45 yield rec
6002cc0df04e planemo upload for repository https://github.com/TAMU-CPT/galaxy-webapollo commit 4e5a5af7689f1713c34a6ad9a9594c205e762fdd
eric-rasche
parents:
diff changeset
46
6002cc0df04e planemo upload for repository https://github.com/TAMU-CPT/galaxy-webapollo commit 4e5a5af7689f1713c34a6ad9a9594c205e762fdd
eric-rasche
parents:
diff changeset
47 if __name__ == '__main__':
6002cc0df04e planemo upload for repository https://github.com/TAMU-CPT/galaxy-webapollo commit 4e5a5af7689f1713c34a6ad9a9594c205e762fdd
eric-rasche
parents:
diff changeset
48 parser = argparse.ArgumentParser(description='rebase gff3 features against parent locations', epilog="")
6002cc0df04e planemo upload for repository https://github.com/TAMU-CPT/galaxy-webapollo commit 4e5a5af7689f1713c34a6ad9a9594c205e762fdd
eric-rasche
parents:
diff changeset
49 parser.add_argument('annotations', type=file, help='Parent GFF3 annotations')
6002cc0df04e planemo upload for repository https://github.com/TAMU-CPT/galaxy-webapollo commit 4e5a5af7689f1713c34a6ad9a9594c205e762fdd
eric-rasche
parents:
diff changeset
50 # parser.add_argument('genome', type=file, help='Genome Sequence')
6002cc0df04e planemo upload for repository https://github.com/TAMU-CPT/galaxy-webapollo commit 4e5a5af7689f1713c34a6ad9a9594c205e762fdd
eric-rasche
parents:
diff changeset
51 parser.add_argument('--user_email')
6002cc0df04e planemo upload for repository https://github.com/TAMU-CPT/galaxy-webapollo commit 4e5a5af7689f1713c34a6ad9a9594c205e762fdd
eric-rasche
parents:
diff changeset
52
6002cc0df04e planemo upload for repository https://github.com/TAMU-CPT/galaxy-webapollo commit 4e5a5af7689f1713c34a6ad9a9594c205e762fdd
eric-rasche
parents:
diff changeset
53 args = parser.parse_args()
6002cc0df04e planemo upload for repository https://github.com/TAMU-CPT/galaxy-webapollo commit 4e5a5af7689f1713c34a6ad9a9594c205e762fdd
eric-rasche
parents:
diff changeset
54 for rec in fix_apollo_issues(**vars(args)):
6002cc0df04e planemo upload for repository https://github.com/TAMU-CPT/galaxy-webapollo commit 4e5a5af7689f1713c34a6ad9a9594c205e762fdd
eric-rasche
parents:
diff changeset
55 rec.annotations = {}
6002cc0df04e planemo upload for repository https://github.com/TAMU-CPT/galaxy-webapollo commit 4e5a5af7689f1713c34a6ad9a9594c205e762fdd
eric-rasche
parents:
diff changeset
56 GFF.write([rec], sys.stdout)