|
0
|
1 #!/usr/bin/env python
|
|
|
2 import argparse
|
|
|
3 import sys
|
|
|
4 import logging
|
|
|
5 from CPT_GFFParser import gffParse, gffWrite
|
|
|
6 from gff3 import feature_lambda, feature_test_type
|
|
|
7
|
|
|
8 logging.basicConfig(level=logging.INFO)
|
|
|
9 log = logging.getLogger(__name__)
|
|
|
10
|
|
|
11
|
|
|
12 def promote_qualifier(qualifier, parent, child, gff3):
|
|
|
13 for record in gffParse(gff3):
|
|
|
14 for parent_feature in feature_lambda(
|
|
|
15 record.features, feature_test_type, {"type": parent}, subfeatures=True
|
|
|
16 ):
|
|
|
17 # for each feature of the parent type, get the first subfeature of the child type
|
|
|
18 try:
|
|
|
19 first_child = sorted(
|
|
|
20 list(
|
|
|
21 feature_lambda(
|
|
|
22 parent_feature.sub_features,
|
|
|
23 feature_test_type,
|
|
|
24 {"type": child},
|
|
|
25 subfeatures=False,
|
|
|
26 )
|
|
|
27 ),
|
|
|
28 key=lambda x: x.location.start
|
|
|
29 if parent_feature.strand > 0
|
|
|
30 else x.location.end,
|
|
|
31 reverse=False if parent_feature.strand > 0 else True,
|
|
|
32 )[0]
|
|
|
33 except IndexError:
|
|
|
34 logging.warning("Child type %s not found under parent %s" % (child, parent_feature.qualifiers["ID"]))
|
|
|
35 continue
|
|
|
36 try:
|
|
|
37 parent_feature.qualifiers[qualifier] = first_child.qualifiers[qualifier]
|
|
|
38 logging.info(
|
|
|
39 "Promoted %s=%s in child %s to parent %s"
|
|
|
40 % (
|
|
|
41 qualifier,
|
|
|
42 first_child.qualifiers[qualifier],
|
|
|
43 first_child.qualifiers["ID"],
|
|
|
44 parent_feature.qualifiers["ID"],
|
|
|
45 )
|
|
|
46 )
|
|
|
47 except KeyError:
|
|
|
48 logging.warning(
|
|
|
49 "Qualifier %s not found in child feature %s"
|
|
|
50 % (qualifier, first_child.qualifiers["ID"])
|
|
|
51 )
|
|
|
52 gffWrite([record], sys.stdout)
|
|
|
53
|
|
|
54
|
|
|
55 if __name__ == "__main__":
|
|
|
56 parser = argparse.ArgumentParser(
|
|
|
57 description="Promote a child feature's qualifer to the parent feature's qualifier",
|
|
|
58 epilog="",
|
|
|
59 )
|
|
|
60 parser.add_argument("gff3", type=argparse.FileType("r"), help="GFF3 File")
|
|
|
61 parser.add_argument(
|
|
|
62 "parent",
|
|
|
63 type=str,
|
|
|
64 help="Feature type of the target parent feature (ex: gene, mrna, exon",
|
|
|
65 )
|
|
|
66 parser.add_argument(
|
|
|
67 "child",
|
|
|
68 type=str,
|
|
|
69 help="Feature type of the target child feature (ex: mrna, exon, CDS",
|
|
|
70 )
|
|
|
71 parser.add_argument(
|
|
|
72 "qualifier", help="Sepcific qualifier to promote (ex: Name, product, notes"
|
|
|
73 )
|
|
|
74 args = parser.parse_args()
|
|
|
75 promote_qualifier(**vars(args))
|
