comparison cpt_starts/start_stats.py @ 0:9f2517655a1e draft

Uploaded
author cpt
date Fri, 13 May 2022 05:38:37 +0000
parents
children
comparison
equal deleted inserted replaced
-1:000000000000 0:9f2517655a1e
1 #!/usr/bin/env python
2 import argparse
3 from CPT_GFFParser import gffParse, gffWrite
4 from Bio import SeqIO
5 from gff3 import feature_lambda, feature_test_type
6
7
8 def main(fasta, gff3):
9 seq_dict = SeqIO.to_dict(SeqIO.parse(fasta, "fasta"))
10
11 codon_usage = {}
12
13 for rec in gffParse(gff3, base_dict=seq_dict):
14 for feat in feature_lambda(
15 rec.features, feature_test_type, {"type": "CDS"}, subfeatures=True
16 ):
17 seq = str(feat.extract(rec).seq)[0:3]
18 try:
19 codon_usage[seq] += 1
20 except KeyError:
21 codon_usage[seq] = 1
22
23 # TODO: print all actg combinations? Or just ones that are there
24 print ("# Codon\tCount")
25 for key in sorted(codon_usage):
26 print ("\t".join((key, str(codon_usage[key]))))
27
28
29 if __name__ == "__main__":
30 parser = argparse.ArgumentParser(
31 description="Summarise start codon usage", epilog=""
32 )
33 parser.add_argument("fasta", type=argparse.FileType("r"), help="Fasta Genome")
34 parser.add_argument("gff3", type=argparse.FileType("r"), help="GFF3 File")
35 args = parser.parse_args()
36 main(**vars(args))