Mercurial > repos > vipints > fml_gff3togtf

annotate bed_to_gff.py @ 10:c42c69aa81f8

Find changesets by keywords (author, files, the commit message), revision number or hash, or revset expression.
fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
author vipints <vipin@cbio.mskcc.org>
date Thu, 23 Apr 2015 18:01:45 -0400
parents
children
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
rev   line source
10
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
1 #!/usr/bin/env python
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
2 """
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
3 Convert genome annotation data in a 12 column BED format to GFF3.
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
4
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
5 Usage:
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
6 python bed_to_gff.py in.bed > out.gff
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
7
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
8 Requirement:
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
9 helper.py : https://github.com/vipints/GFFtools-GX/blob/master/helper.py
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
10
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
11 Copyright (C)
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
12 2009-2012 Friedrich Miescher Laboratory of the Max Planck Society, Tubingen, Germany.
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
13 2012-2015 Memorial Sloan Kettering Cancer Center New York City, USA.
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
14 """
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
15
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
16 import re
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
17 import sys
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
18 import helper
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
19
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
20 def __main__():
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
21 """
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
22 main function
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
23 """
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
24
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
25 try:
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
26 bed_fname = sys.argv[1]
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
27 except:
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
28 print __doc__
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
29 sys.exit(-1)
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
30
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
31 bed_fh = helper.open_file(bed_fname)
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
32
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
33 for line in bed_fh:
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
34 line = line.strip( '\n\r' )
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
35
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
36 if not line or line[0] in ['#']:
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
37 continue
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
38
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
39 parts = line.split('\t')
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
40 assert len(parts) >= 12, line
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
41
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
42 rstarts = parts[-1].split(',')
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
43 rstarts.pop() if rstarts[-1] == '' else rstarts
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
44
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
45 exon_lens = parts[-2].split(',')
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
46 exon_lens.pop() if exon_lens[-1] == '' else exon_lens
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
47
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
48 if len(rstarts) != len(exon_lens):
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
49 continue # checking the consistency col 11 and col 12
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
50
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
51 if len(rstarts) != int(parts[-3]):
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
52 continue # checking the number of exons and block count are same
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
53
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
54 if not parts[5] in ['+', '-']:
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
55 parts[5] = '.' # replace the unknown strand with '.'
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
56
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
57 # bed2gff result line
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
58 sys.stdout.write('%s\tbed2gff\tgene\t%d\t%s\t%s\t%s\t.\tID=Gene:%s;Name=Gene:%s\n' % (parts[0], int(parts[1])+1, parts[2], parts[4], parts[5], parts[3], parts[3]))
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
59 sys.stdout.write('%s\tbed2gff\ttranscript\t%d\t%s\t%s\t%s\t.\tID=%s;Name=%s;Parent=Gene:%s\n' % (parts[0], int(parts[1])+1, parts[2], parts[4], parts[5], parts[3], parts[3], parts[3]))
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
60
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
61 st = int(parts[1])
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
62 for ex_cnt in range(int(parts[-3])):
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
63 start = st + int(rstarts[ex_cnt]) + 1
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
64 stop = start + int(exon_lens[ex_cnt]) - 1
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
65 sys.stdout.write('%s\tbed2gff\texon\t%d\t%d\t%s\t%s\t.\tParent=%s\n' % (parts[0], start, stop, parts[4], parts[5], parts[3]))
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
66
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
67 bed_fh.close()
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
68
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
69
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
70 if __name__ == "__main__":
c42c69aa81f8 fixed manually the upload of version 2.1.0 - deleted accidentally added files to the repo
vipints <vipin@cbio.mskcc.org>
parents:
diff changeset
71 __main__()