Mercurial > repos > artbio > rsem

annotate purge_gtf_from_multichrom_genes.py @ 11:c86ed39b72eb draft

Find changesets by keywords (author, files, the commit message), revision number or hash, or revset expression.
"planemo upload for repository https://github.com/artbio/tools-artbio/tree/master/tools/rsem commit 10719ebe96c608be124360a4be62ec4d164412b3"
author artbio
date Sun, 05 Jan 2020 22:07:36 +0000
parents
children
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
rev   line source
11
c86ed39b72eb "planemo upload for repository https://github.com/artbio/tools-artbio/tree/master/tools/rsem commit 10719ebe96c608be124360a4be62ec4d164412b3"
artbio
parents:
diff changeset
1 #!/usr/bin/env python
c86ed39b72eb "planemo upload for repository https://github.com/artbio/tools-artbio/tree/master/tools/rsem commit 10719ebe96c608be124360a4be62ec4d164412b3"
artbio
parents:
diff changeset
2
c86ed39b72eb "planemo upload for repository https://github.com/artbio/tools-artbio/tree/master/tools/rsem commit 10719ebe96c608be124360a4be62ec4d164412b3"
artbio
parents:
diff changeset
3 import argparse
c86ed39b72eb "planemo upload for repository https://github.com/artbio/tools-artbio/tree/master/tools/rsem commit 10719ebe96c608be124360a4be62ec4d164412b3"
artbio
parents:
diff changeset
4 from collections import defaultdict
c86ed39b72eb "planemo upload for repository https://github.com/artbio/tools-artbio/tree/master/tools/rsem commit 10719ebe96c608be124360a4be62ec4d164412b3"
artbio
parents:
diff changeset
5
c86ed39b72eb "planemo upload for repository https://github.com/artbio/tools-artbio/tree/master/tools/rsem commit 10719ebe96c608be124360a4be62ec4d164412b3"
artbio
parents:
diff changeset
6
c86ed39b72eb "planemo upload for repository https://github.com/artbio/tools-artbio/tree/master/tools/rsem commit 10719ebe96c608be124360a4be62ec4d164412b3"
artbio
parents:
diff changeset
7 def command_parse():
c86ed39b72eb "planemo upload for repository https://github.com/artbio/tools-artbio/tree/master/tools/rsem commit 10719ebe96c608be124360a4be62ec4d164412b3"
artbio
parents:
diff changeset
8 parser = argparse.ArgumentParser(description='Purge GTF file from genes \
c86ed39b72eb "planemo upload for repository https://github.com/artbio/tools-artbio/tree/master/tools/rsem commit 10719ebe96c608be124360a4be62ec4d164412b3"
artbio
parents:
diff changeset
9 that are on several chromosomes and list them in a log file')
c86ed39b72eb "planemo upload for repository https://github.com/artbio/tools-artbio/tree/master/tools/rsem commit 10719ebe96c608be124360a4be62ec4d164412b3"
artbio
parents:
diff changeset
10 parser.add_argument(
c86ed39b72eb "planemo upload for repository https://github.com/artbio/tools-artbio/tree/master/tools/rsem commit 10719ebe96c608be124360a4be62ec4d164412b3"
artbio
parents:
diff changeset
11 '-i', '--input', dest='input', help='input GTF file', required=True)
c86ed39b72eb "planemo upload for repository https://github.com/artbio/tools-artbio/tree/master/tools/rsem commit 10719ebe96c608be124360a4be62ec4d164412b3"
artbio
parents:
diff changeset
12 parser.add_argument('-o', '--output', dest='output', help='output file \
c86ed39b72eb "planemo upload for repository https://github.com/artbio/tools-artbio/tree/master/tools/rsem commit 10719ebe96c608be124360a4be62ec4d164412b3"
artbio
parents:
diff changeset
13 name', default='output.gtf')
c86ed39b72eb "planemo upload for repository https://github.com/artbio/tools-artbio/tree/master/tools/rsem commit 10719ebe96c608be124360a4be62ec4d164412b3"
artbio
parents:
diff changeset
14 parser.add_argument('-l', '--log', dest='log', help='log of purged \
c86ed39b72eb "planemo upload for repository https://github.com/artbio/tools-artbio/tree/master/tools/rsem commit 10719ebe96c608be124360a4be62ec4d164412b3"
artbio
parents:
diff changeset
15 genes', default='purged_genes.log')
c86ed39b72eb "planemo upload for repository https://github.com/artbio/tools-artbio/tree/master/tools/rsem commit 10719ebe96c608be124360a4be62ec4d164412b3"
artbio
parents:
diff changeset
16 args = parser.parse_args()
c86ed39b72eb "planemo upload for repository https://github.com/artbio/tools-artbio/tree/master/tools/rsem commit 10719ebe96c608be124360a4be62ec4d164412b3"
artbio
parents:
diff changeset
17 return args
c86ed39b72eb "planemo upload for repository https://github.com/artbio/tools-artbio/tree/master/tools/rsem commit 10719ebe96c608be124360a4be62ec4d164412b3"
artbio
parents:
diff changeset
18
c86ed39b72eb "planemo upload for repository https://github.com/artbio/tools-artbio/tree/master/tools/rsem commit 10719ebe96c608be124360a4be62ec4d164412b3"
artbio
parents:
diff changeset
19
c86ed39b72eb "planemo upload for repository https://github.com/artbio/tools-artbio/tree/master/tools/rsem commit 10719ebe96c608be124360a4be62ec4d164412b3"
artbio
parents:
diff changeset
20 def get_genes(gtf_file):
c86ed39b72eb "planemo upload for repository https://github.com/artbio/tools-artbio/tree/master/tools/rsem commit 10719ebe96c608be124360a4be62ec4d164412b3"
artbio
parents:
diff changeset
21 genes = defaultdict(list)
c86ed39b72eb "planemo upload for repository https://github.com/artbio/tools-artbio/tree/master/tools/rsem commit 10719ebe96c608be124360a4be62ec4d164412b3"
artbio
parents:
diff changeset
22 with open(gtf_file, 'r') as fh:
c86ed39b72eb "planemo upload for repository https://github.com/artbio/tools-artbio/tree/master/tools/rsem commit 10719ebe96c608be124360a4be62ec4d164412b3"
artbio
parents:
diff changeset
23 for line in fh:
c86ed39b72eb "planemo upload for repository https://github.com/artbio/tools-artbio/tree/master/tools/rsem commit 10719ebe96c608be124360a4be62ec4d164412b3"
artbio
parents:
diff changeset
24 if line[0] != '#':
c86ed39b72eb "planemo upload for repository https://github.com/artbio/tools-artbio/tree/master/tools/rsem commit 10719ebe96c608be124360a4be62ec4d164412b3"
artbio
parents:
diff changeset
25 fields = line[:-1].split("\t")
c86ed39b72eb "planemo upload for repository https://github.com/artbio/tools-artbio/tree/master/tools/rsem commit 10719ebe96c608be124360a4be62ec4d164412b3"
artbio
parents:
diff changeset
26 chrom = fields[0]
c86ed39b72eb "planemo upload for repository https://github.com/artbio/tools-artbio/tree/master/tools/rsem commit 10719ebe96c608be124360a4be62ec4d164412b3"
artbio
parents:
diff changeset
27 name_gene = fields[-1].split('gene_id "')[-1].split('"; \
c86ed39b72eb "planemo upload for repository https://github.com/artbio/tools-artbio/tree/master/tools/rsem commit 10719ebe96c608be124360a4be62ec4d164412b3"
artbio
parents:
diff changeset
28 transcript_id')[0]
c86ed39b72eb "planemo upload for repository https://github.com/artbio/tools-artbio/tree/master/tools/rsem commit 10719ebe96c608be124360a4be62ec4d164412b3"
artbio
parents:
diff changeset
29 genes[name_gene].append(chrom)
c86ed39b72eb "planemo upload for repository https://github.com/artbio/tools-artbio/tree/master/tools/rsem commit 10719ebe96c608be124360a4be62ec4d164412b3"
artbio
parents:
diff changeset
30 return genes
c86ed39b72eb "planemo upload for repository https://github.com/artbio/tools-artbio/tree/master/tools/rsem commit 10719ebe96c608be124360a4be62ec4d164412b3"
artbio
parents:
diff changeset
31
c86ed39b72eb "planemo upload for repository https://github.com/artbio/tools-artbio/tree/master/tools/rsem commit 10719ebe96c608be124360a4be62ec4d164412b3"
artbio
parents:
diff changeset
32
c86ed39b72eb "planemo upload for repository https://github.com/artbio/tools-artbio/tree/master/tools/rsem commit 10719ebe96c608be124360a4be62ec4d164412b3"
artbio
parents:
diff changeset
33 def generate_output(genes, log_file):
c86ed39b72eb "planemo upload for repository https://github.com/artbio/tools-artbio/tree/master/tools/rsem commit 10719ebe96c608be124360a4be62ec4d164412b3"
artbio
parents:
diff changeset
34 '''
c86ed39b72eb "planemo upload for repository https://github.com/artbio/tools-artbio/tree/master/tools/rsem commit 10719ebe96c608be124360a4be62ec4d164412b3"
artbio
parents:
diff changeset
35 Search for all genes that are present on several chromosomes. This function
c86ed39b72eb "planemo upload for repository https://github.com/artbio/tools-artbio/tree/master/tools/rsem commit 10719ebe96c608be124360a4be62ec4d164412b3"
artbio
parents:
diff changeset
36 return a list of these genes in target_genes. It also generate a log tab
c86ed39b72eb "planemo upload for repository https://github.com/artbio/tools-artbio/tree/master/tools/rsem commit 10719ebe96c608be124360a4be62ec4d164412b3"
artbio
parents:
diff changeset
37 delimited file with one gene per line and with its list of chromosomes
c86ed39b72eb "planemo upload for repository https://github.com/artbio/tools-artbio/tree/master/tools/rsem commit 10719ebe96c608be124360a4be62ec4d164412b3"
artbio
parents:
diff changeset
38 (coma delimited)
c86ed39b72eb "planemo upload for repository https://github.com/artbio/tools-artbio/tree/master/tools/rsem commit 10719ebe96c608be124360a4be62ec4d164412b3"
artbio
parents:
diff changeset
39 '''
c86ed39b72eb "planemo upload for repository https://github.com/artbio/tools-artbio/tree/master/tools/rsem commit 10719ebe96c608be124360a4be62ec4d164412b3"
artbio
parents:
diff changeset
40 output = open(log_file, 'w')
c86ed39b72eb "planemo upload for repository https://github.com/artbio/tools-artbio/tree/master/tools/rsem commit 10719ebe96c608be124360a4be62ec4d164412b3"
artbio
parents:
diff changeset
41 # output.write('#all genes on several chromosomes' + '\n')
c86ed39b72eb "planemo upload for repository https://github.com/artbio/tools-artbio/tree/master/tools/rsem commit 10719ebe96c608be124360a4be62ec4d164412b3"
artbio
parents:
diff changeset
42 target_genes = list()
c86ed39b72eb "planemo upload for repository https://github.com/artbio/tools-artbio/tree/master/tools/rsem commit 10719ebe96c608be124360a4be62ec4d164412b3"
artbio
parents:
diff changeset
43 for name_gene in genes.keys():
c86ed39b72eb "planemo upload for repository https://github.com/artbio/tools-artbio/tree/master/tools/rsem commit 10719ebe96c608be124360a4be62ec4d164412b3"
artbio
parents:
diff changeset
44 genes[name_gene] = set(genes[name_gene])
c86ed39b72eb "planemo upload for repository https://github.com/artbio/tools-artbio/tree/master/tools/rsem commit 10719ebe96c608be124360a4be62ec4d164412b3"
artbio
parents:
diff changeset
45 if len(genes[name_gene]) > 1:
c86ed39b72eb "planemo upload for repository https://github.com/artbio/tools-artbio/tree/master/tools/rsem commit 10719ebe96c608be124360a4be62ec4d164412b3"
artbio
parents:
diff changeset
46 target_genes.append(name_gene)
c86ed39b72eb "planemo upload for repository https://github.com/artbio/tools-artbio/tree/master/tools/rsem commit 10719ebe96c608be124360a4be62ec4d164412b3"
artbio
parents:
diff changeset
47 new_line = '\t'.join([name_gene, ','.join(genes[name_gene])])
c86ed39b72eb "planemo upload for repository https://github.com/artbio/tools-artbio/tree/master/tools/rsem commit 10719ebe96c608be124360a4be62ec4d164412b3"
artbio
parents:
diff changeset
48 output.write("%s\n" % new_line)
c86ed39b72eb "planemo upload for repository https://github.com/artbio/tools-artbio/tree/master/tools/rsem commit 10719ebe96c608be124360a4be62ec4d164412b3"
artbio
parents:
diff changeset
49 output.close()
c86ed39b72eb "planemo upload for repository https://github.com/artbio/tools-artbio/tree/master/tools/rsem commit 10719ebe96c608be124360a4be62ec4d164412b3"
artbio
parents:
diff changeset
50 return target_genes
c86ed39b72eb "planemo upload for repository https://github.com/artbio/tools-artbio/tree/master/tools/rsem commit 10719ebe96c608be124360a4be62ec4d164412b3"
artbio
parents:
diff changeset
51
c86ed39b72eb "planemo upload for repository https://github.com/artbio/tools-artbio/tree/master/tools/rsem commit 10719ebe96c608be124360a4be62ec4d164412b3"
artbio
parents:
diff changeset
52
c86ed39b72eb "planemo upload for repository https://github.com/artbio/tools-artbio/tree/master/tools/rsem commit 10719ebe96c608be124360a4be62ec4d164412b3"
artbio
parents:
diff changeset
53 def purge_gtf(target_genes, gtf_file, output_file):
c86ed39b72eb "planemo upload for repository https://github.com/artbio/tools-artbio/tree/master/tools/rsem commit 10719ebe96c608be124360a4be62ec4d164412b3"
artbio
parents:
diff changeset
54 '''
c86ed39b72eb "planemo upload for repository https://github.com/artbio/tools-artbio/tree/master/tools/rsem commit 10719ebe96c608be124360a4be62ec4d164412b3"
artbio
parents:
diff changeset
55 Remove all lines of the gtf file where the gene_id is gene of target_genes
c86ed39b72eb "planemo upload for repository https://github.com/artbio/tools-artbio/tree/master/tools/rsem commit 10719ebe96c608be124360a4be62ec4d164412b3"
artbio
parents:
diff changeset
56 list.
c86ed39b72eb "planemo upload for repository https://github.com/artbio/tools-artbio/tree/master/tools/rsem commit 10719ebe96c608be124360a4be62ec4d164412b3"
artbio
parents:
diff changeset
57 '''
c86ed39b72eb "planemo upload for repository https://github.com/artbio/tools-artbio/tree/master/tools/rsem commit 10719ebe96c608be124360a4be62ec4d164412b3"
artbio
parents:
diff changeset
58 output_gtf = open(output_file, 'w')
c86ed39b72eb "planemo upload for repository https://github.com/artbio/tools-artbio/tree/master/tools/rsem commit 10719ebe96c608be124360a4be62ec4d164412b3"
artbio
parents:
diff changeset
59 with open(gtf_file, 'r') as gtf_handler:
c86ed39b72eb "planemo upload for repository https://github.com/artbio/tools-artbio/tree/master/tools/rsem commit 10719ebe96c608be124360a4be62ec4d164412b3"
artbio
parents:
diff changeset
60 for line in gtf_handler:
c86ed39b72eb "planemo upload for repository https://github.com/artbio/tools-artbio/tree/master/tools/rsem commit 10719ebe96c608be124360a4be62ec4d164412b3"
artbio
parents:
diff changeset
61 fields = line[:-1].split("\t")
c86ed39b72eb "planemo upload for repository https://github.com/artbio/tools-artbio/tree/master/tools/rsem commit 10719ebe96c608be124360a4be62ec4d164412b3"
artbio
parents:
diff changeset
62 gene_name = fields[-1].split('gene_id "')[-1].split('"; \
c86ed39b72eb "planemo upload for repository https://github.com/artbio/tools-artbio/tree/master/tools/rsem commit 10719ebe96c608be124360a4be62ec4d164412b3"
artbio
parents:
diff changeset
63 transcript_id')[0]
c86ed39b72eb "planemo upload for repository https://github.com/artbio/tools-artbio/tree/master/tools/rsem commit 10719ebe96c608be124360a4be62ec4d164412b3"
artbio
parents:
diff changeset
64 if gene_name not in target_genes:
c86ed39b72eb "planemo upload for repository https://github.com/artbio/tools-artbio/tree/master/tools/rsem commit 10719ebe96c608be124360a4be62ec4d164412b3"
artbio
parents:
diff changeset
65 output_gtf.write(line)
c86ed39b72eb "planemo upload for repository https://github.com/artbio/tools-artbio/tree/master/tools/rsem commit 10719ebe96c608be124360a4be62ec4d164412b3"
artbio
parents:
diff changeset
66 output_gtf.close()
c86ed39b72eb "planemo upload for repository https://github.com/artbio/tools-artbio/tree/master/tools/rsem commit 10719ebe96c608be124360a4be62ec4d164412b3"
artbio
parents:
diff changeset
67
c86ed39b72eb "planemo upload for repository https://github.com/artbio/tools-artbio/tree/master/tools/rsem commit 10719ebe96c608be124360a4be62ec4d164412b3"
artbio
parents:
diff changeset
68
c86ed39b72eb "planemo upload for repository https://github.com/artbio/tools-artbio/tree/master/tools/rsem commit 10719ebe96c608be124360a4be62ec4d164412b3"
artbio
parents:
diff changeset
69 def __main__():
c86ed39b72eb "planemo upload for repository https://github.com/artbio/tools-artbio/tree/master/tools/rsem commit 10719ebe96c608be124360a4be62ec4d164412b3"
artbio
parents:
diff changeset
70 args = command_parse()
c86ed39b72eb "planemo upload for repository https://github.com/artbio/tools-artbio/tree/master/tools/rsem commit 10719ebe96c608be124360a4be62ec4d164412b3"
artbio
parents:
diff changeset
71 genes = get_genes(args.input)
c86ed39b72eb "planemo upload for repository https://github.com/artbio/tools-artbio/tree/master/tools/rsem commit 10719ebe96c608be124360a4be62ec4d164412b3"
artbio
parents:
diff changeset
72 target_genes = generate_output(genes, args.log)
c86ed39b72eb "planemo upload for repository https://github.com/artbio/tools-artbio/tree/master/tools/rsem commit 10719ebe96c608be124360a4be62ec4d164412b3"
artbio
parents:
diff changeset
73 purge_gtf(target_genes, args.input, args.output)
c86ed39b72eb "planemo upload for repository https://github.com/artbio/tools-artbio/tree/master/tools/rsem commit 10719ebe96c608be124360a4be62ec4d164412b3"
artbio
parents:
diff changeset
74
c86ed39b72eb "planemo upload for repository https://github.com/artbio/tools-artbio/tree/master/tools/rsem commit 10719ebe96c608be124360a4be62ec4d164412b3"
artbio
parents:
diff changeset
75
c86ed39b72eb "planemo upload for repository https://github.com/artbio/tools-artbio/tree/master/tools/rsem commit 10719ebe96c608be124360a4be62ec4d164412b3"
artbio
parents:
diff changeset
76 if __name__ == "__main__":
c86ed39b72eb "planemo upload for repository https://github.com/artbio/tools-artbio/tree/master/tools/rsem commit 10719ebe96c608be124360a4be62ec4d164412b3"
artbio
parents:
diff changeset
77 __main__()