Mercurial > repos > cpt > cpt_gff_add_parents
comparison cpt_gff_add_parent/gff3_add_parents_to_cds.py @ 0:60651e90119a draft
Uploaded
author | cpt |
---|---|
date | Fri, 13 May 2022 04:54:17 +0000 |
parents | |
children |
comparison
equal
deleted
inserted
replaced
-1:000000000000 | 0:60651e90119a |
---|---|
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 feature in feature_lambda( | |
15 rec.features, feature_test_type, {"type": "CDS"}, subfeatures=True | |
16 ): | |
17 import random | |
18 | |
19 fid = feature.qualifiers["ID"][0] + "_" + str(random.random()) | |
20 gene = gffSeqFeature( | |
21 location=feature.location, | |
22 type="gene", | |
23 qualifiers={"ID": [fid], "source": ["cpt.fixModel"]}, | |
24 ) | |
25 # Below that we have an mRNA | |
26 mRNA = gffSeqFeature( | |
27 location=feature.location, | |
28 type="mRNA", | |
29 qualifiers={"source": ["cpt.fixModel"], "ID": ["%s.mRNA" % fid], "Parent": gene.qualifiers["ID"]}, | |
30 ) | |
31 feature.qualifiers["ID"] = [fid + ".CDS"] | |
32 feature.qualifiers["Parent"] = mRNA.qualifiers["ID"] | |
33 | |
34 mRNA.sub_features = [feature] | |
35 gene.sub_features = [mRNA] | |
36 yield gene | |
37 | |
38 | |
39 def gff_filter(gff3): | |
40 for rec in gffParse(gff3): | |
41 rec.features = sorted(list(fixed_feature(rec)), key=lambda x: x.location.start) | |
42 rec.annotations = {} | |
43 gffWrite([rec], sys.stdout) | |
44 | |
45 | |
46 if __name__ == "__main__": | |
47 parser = argparse.ArgumentParser(description="add parent gene features to CDSs") | |
48 parser.add_argument("gff3", type=argparse.FileType("r"), help="GFF3 annotations") | |
49 args = parser.parse_args() | |
50 gff_filter(**vars(args)) |