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

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