annotate gff3_add_parents_to_cds.py @ 0:b6ff405a3481 draft default tip

Uploaded
author cpt_testbed
date Fri, 29 Apr 2022 11:21:36 +0000
parents
children
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
rev   line source
0
b6ff405a3481 Uploaded
cpt_testbed
parents:
diff changeset
1 #!/usr/bin/env python
b6ff405a3481 Uploaded
cpt_testbed
parents:
diff changeset
2 import sys
b6ff405a3481 Uploaded
cpt_testbed
parents:
diff changeset
3 import logging
b6ff405a3481 Uploaded
cpt_testbed
parents:
diff changeset
4 import argparse
b6ff405a3481 Uploaded
cpt_testbed
parents:
diff changeset
5 from CPT_GFFParser import gffParse, gffWrite, gffSeqFeature
b6ff405a3481 Uploaded
cpt_testbed
parents:
diff changeset
6 #from Bio.SeqFeature import SeqFeature
b6ff405a3481 Uploaded
cpt_testbed
parents:
diff changeset
7 from gff3 import feature_lambda, feature_test_type
b6ff405a3481 Uploaded
cpt_testbed
parents:
diff changeset
8
b6ff405a3481 Uploaded
cpt_testbed
parents:
diff changeset
9 logging.basicConfig(level=logging.INFO)
b6ff405a3481 Uploaded
cpt_testbed
parents:
diff changeset
10 log = logging.getLogger(__name__)
b6ff405a3481 Uploaded
cpt_testbed
parents:
diff changeset
11
b6ff405a3481 Uploaded
cpt_testbed
parents:
diff changeset
12
b6ff405a3481 Uploaded
cpt_testbed
parents:
diff changeset
13 def fixed_feature(rec):
b6ff405a3481 Uploaded
cpt_testbed
parents:
diff changeset
14 for feature in feature_lambda(
b6ff405a3481 Uploaded
cpt_testbed
parents:
diff changeset
15 rec.features, feature_test_type, {"type": "CDS"}, subfeatures=True
b6ff405a3481 Uploaded
cpt_testbed
parents:
diff changeset
16 ):
b6ff405a3481 Uploaded
cpt_testbed
parents:
diff changeset
17 import random
b6ff405a3481 Uploaded
cpt_testbed
parents:
diff changeset
18
b6ff405a3481 Uploaded
cpt_testbed
parents:
diff changeset
19 fid = feature.qualifiers["ID"][0] + "_" + str(random.random())
b6ff405a3481 Uploaded
cpt_testbed
parents:
diff changeset
20 gene = gffSeqFeature(
b6ff405a3481 Uploaded
cpt_testbed
parents:
diff changeset
21 location=feature.location,
b6ff405a3481 Uploaded
cpt_testbed
parents:
diff changeset
22 type="gene",
b6ff405a3481 Uploaded
cpt_testbed
parents:
diff changeset
23 qualifiers={"ID": [fid], "source": ["cpt.fixModel"]},
b6ff405a3481 Uploaded
cpt_testbed
parents:
diff changeset
24 )
b6ff405a3481 Uploaded
cpt_testbed
parents:
diff changeset
25 # Below that we have an mRNA
b6ff405a3481 Uploaded
cpt_testbed
parents:
diff changeset
26 mRNA = gffSeqFeature(
b6ff405a3481 Uploaded
cpt_testbed
parents:
diff changeset
27 location=feature.location,
b6ff405a3481 Uploaded
cpt_testbed
parents:
diff changeset
28 type="mRNA",
b6ff405a3481 Uploaded
cpt_testbed
parents:
diff changeset
29 qualifiers={"source": ["cpt.fixModel"], "ID": ["%s.mRNA" % fid], "Parent": gene.qualifiers["ID"]},
b6ff405a3481 Uploaded
cpt_testbed
parents:
diff changeset
30 )
b6ff405a3481 Uploaded
cpt_testbed
parents:
diff changeset
31 feature.qualifiers["ID"] = [fid + ".CDS"]
b6ff405a3481 Uploaded
cpt_testbed
parents:
diff changeset
32 feature.qualifiers["Parent"] = mRNA.qualifiers["ID"]
b6ff405a3481 Uploaded
cpt_testbed
parents:
diff changeset
33
b6ff405a3481 Uploaded
cpt_testbed
parents:
diff changeset
34 mRNA.sub_features = [feature]
b6ff405a3481 Uploaded
cpt_testbed
parents:
diff changeset
35 gene.sub_features = [mRNA]
b6ff405a3481 Uploaded
cpt_testbed
parents:
diff changeset
36 yield gene
b6ff405a3481 Uploaded
cpt_testbed
parents:
diff changeset
37
b6ff405a3481 Uploaded
cpt_testbed
parents:
diff changeset
38
b6ff405a3481 Uploaded
cpt_testbed
parents:
diff changeset
39 def gff_filter(gff3):
b6ff405a3481 Uploaded
cpt_testbed
parents:
diff changeset
40 for rec in gffParse(gff3):
b6ff405a3481 Uploaded
cpt_testbed
parents:
diff changeset
41 rec.features = sorted(list(fixed_feature(rec)), key=lambda x: x.location.start)
b6ff405a3481 Uploaded
cpt_testbed
parents:
diff changeset
42 rec.annotations = {}
b6ff405a3481 Uploaded
cpt_testbed
parents:
diff changeset
43 gffWrite([rec], sys.stdout)
b6ff405a3481 Uploaded
cpt_testbed
parents:
diff changeset
44
b6ff405a3481 Uploaded
cpt_testbed
parents:
diff changeset
45
b6ff405a3481 Uploaded
cpt_testbed
parents:
diff changeset
46 if __name__ == "__main__":
b6ff405a3481 Uploaded
cpt_testbed
parents:
diff changeset
47 parser = argparse.ArgumentParser(description="add parent gene features to CDSs")
b6ff405a3481 Uploaded
cpt_testbed
parents:
diff changeset
48 parser.add_argument("gff3", type=argparse.FileType("r"), help="GFF3 annotations")
b6ff405a3481 Uploaded
cpt_testbed
parents:
diff changeset
49 args = parser.parse_args()
b6ff405a3481 Uploaded
cpt_testbed
parents:
diff changeset
50 gff_filter(**vars(args))