comparison fix-aragorn-gff3.py @ 0:d8752390c7f7 draft default tip

Uploaded
author cpt_testbed
date Fri, 29 Apr 2022 11:32:30 +0000
parents
children
comparison
equal deleted inserted replaced
-1:000000000000 0:d8752390c7f7
1 #!/usr/bin/env python
2 import sys
3 import logging
4 import argparse
5 from CPT_GFFParser import gffParse, gffWrite, gffSeqFeature
6 from Bio.SeqFeature import SeqFeature
7 from gff3 import feature_lambda, feature_test_type
8
9 logging.basicConfig(level=logging.INFO)
10 log = logging.getLogger(__name__)
11
12
13 def fixed_feature(rec):
14 for idx, feature in enumerate(
15 feature_lambda(
16 rec.features, feature_test_type, {"types": ["tRNA", "tmRNA"]}, subfeatures=True
17 )
18 ):
19
20 fid = "%s-%03d" % (feature.type, 1 + idx)
21 try:
22 name = [feature.type + "-" + feature.qualifiers["Codon"][0]]
23 except KeyError:
24 name = [feature.qualifiers['product'][0]]
25 try:
26 origSource = feature.qualifiers["source"][0]
27 except:
28 origSource = "."
29 gene = gffSeqFeature(
30 location=feature.location,
31 type="gene",
32 qualifiers={"ID": [fid + ".gene"], "source": [origSource], "Name": name},
33 )
34 feature.qualifiers["Name"] = name
35 # Below that we have an mRNA
36 exon = gffSeqFeature(
37 location=feature.location,
38 type="exon",
39 qualifiers={"source": [origSource], "ID": ["%s.exon" % fid], "Name": name},
40 )
41 feature.qualifiers["ID"] = [fid]
42 exon.qualifiers["Parent"] = [fid]
43 feature.qualifiers["Parent"] = [fid + ".gene"]
44 # gene -> trna -> exon
45 feature.sub_features = [exon]
46 gene.sub_features = [feature]
47 yield gene
48
49
50 def gff_filter(gff3):
51 found_gff = False
52 for rec in gffParse(gff3):
53 found_gff = True
54 rec.features = sorted(list(fixed_feature(rec)), key=lambda x: x.location.start)
55 rec.annotations = {}
56 gffWrite([rec], sys.stdout)
57 if not found_gff:
58 print("##gff-version 3")
59
60
61 if __name__ == "__main__":
62 parser = argparse.ArgumentParser(description="add parent gene features to CDSs")
63 parser.add_argument("gff3", type=argparse.FileType("r"), help="GFF3 annotations")
64 args = parser.parse_args()
65 gff_filter(**vars(args))