Mercurial > repos > galaxyp > retrieve_ensembl_bed

annotate retrieve_ensembl_bed.py @ 1:9c4a48f5d4e7 draft default tip

Find changesets by keywords (author, files, the commit message), revision number or hash, or revset expression.
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/proteogenomics/retrieve_ensembl_bed commit 6babd357845126292cb202aaea0f70ff68819525"
author galaxyp
date Mon, 07 Oct 2019 16:14:39 -0400
parents da1b538b87e5
children
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
rev   line source
0
da1b538b87e5 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/proteogenomics/retrieve_ensembl_bed commit 88cf1e923a8c9e5bc6953ad412d15a7c70f054d1
galaxyp
parents:
diff changeset
1 #!/usr/bin/env python
da1b538b87e5 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/proteogenomics/retrieve_ensembl_bed commit 88cf1e923a8c9e5bc6953ad412d15a7c70f054d1
galaxyp
parents:
diff changeset
2 """
da1b538b87e5 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/proteogenomics/retrieve_ensembl_bed commit 88cf1e923a8c9e5bc6953ad412d15a7c70f054d1
galaxyp
parents:
diff changeset
3 #
da1b538b87e5 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/proteogenomics/retrieve_ensembl_bed commit 88cf1e923a8c9e5bc6953ad412d15a7c70f054d1
galaxyp
parents:
diff changeset
4 #------------------------------------------------------------------------------
da1b538b87e5 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/proteogenomics/retrieve_ensembl_bed commit 88cf1e923a8c9e5bc6953ad412d15a7c70f054d1
galaxyp
parents:
diff changeset
5 # University of Minnesota
da1b538b87e5 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/proteogenomics/retrieve_ensembl_bed commit 88cf1e923a8c9e5bc6953ad412d15a7c70f054d1
galaxyp
parents:
diff changeset
6 # Copyright 2017, Regents of the University of Minnesota
da1b538b87e5 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/proteogenomics/retrieve_ensembl_bed commit 88cf1e923a8c9e5bc6953ad412d15a7c70f054d1
galaxyp
parents:
diff changeset
7 #------------------------------------------------------------------------------
da1b538b87e5 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/proteogenomics/retrieve_ensembl_bed commit 88cf1e923a8c9e5bc6953ad412d15a7c70f054d1
galaxyp
parents:
diff changeset
8 # Author:
da1b538b87e5 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/proteogenomics/retrieve_ensembl_bed commit 88cf1e923a8c9e5bc6953ad412d15a7c70f054d1
galaxyp
parents:
diff changeset
9 #
da1b538b87e5 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/proteogenomics/retrieve_ensembl_bed commit 88cf1e923a8c9e5bc6953ad412d15a7c70f054d1
galaxyp
parents:
diff changeset
10 # James E Johnson
da1b538b87e5 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/proteogenomics/retrieve_ensembl_bed commit 88cf1e923a8c9e5bc6953ad412d15a7c70f054d1
galaxyp
parents:
diff changeset
11 #
da1b538b87e5 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/proteogenomics/retrieve_ensembl_bed commit 88cf1e923a8c9e5bc6953ad412d15a7c70f054d1
galaxyp
parents:
diff changeset
12 #------------------------------------------------------------------------------
da1b538b87e5 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/proteogenomics/retrieve_ensembl_bed commit 88cf1e923a8c9e5bc6953ad412d15a7c70f054d1
galaxyp
parents:
diff changeset
13 """
da1b538b87e5 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/proteogenomics/retrieve_ensembl_bed commit 88cf1e923a8c9e5bc6953ad412d15a7c70f054d1
galaxyp
parents:
diff changeset
14
da1b538b87e5 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/proteogenomics/retrieve_ensembl_bed commit 88cf1e923a8c9e5bc6953ad412d15a7c70f054d1
galaxyp
parents:
diff changeset
15 from __future__ import print_function
da1b538b87e5 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/proteogenomics/retrieve_ensembl_bed commit 88cf1e923a8c9e5bc6953ad412d15a7c70f054d1
galaxyp
parents:
diff changeset
16
da1b538b87e5 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/proteogenomics/retrieve_ensembl_bed commit 88cf1e923a8c9e5bc6953ad412d15a7c70f054d1
galaxyp
parents:
diff changeset
17 import argparse
da1b538b87e5 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/proteogenomics/retrieve_ensembl_bed commit 88cf1e923a8c9e5bc6953ad412d15a7c70f054d1
galaxyp
parents:
diff changeset
18 import re
da1b538b87e5 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/proteogenomics/retrieve_ensembl_bed commit 88cf1e923a8c9e5bc6953ad412d15a7c70f054d1
galaxyp
parents:
diff changeset
19 import sys
da1b538b87e5 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/proteogenomics/retrieve_ensembl_bed commit 88cf1e923a8c9e5bc6953ad412d15a7c70f054d1
galaxyp
parents:
diff changeset
20
da1b538b87e5 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/proteogenomics/retrieve_ensembl_bed commit 88cf1e923a8c9e5bc6953ad412d15a7c70f054d1
galaxyp
parents:
diff changeset
21 from bedutil import bed_from_line
da1b538b87e5 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/proteogenomics/retrieve_ensembl_bed commit 88cf1e923a8c9e5bc6953ad412d15a7c70f054d1
galaxyp
parents:
diff changeset
22
da1b538b87e5 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/proteogenomics/retrieve_ensembl_bed commit 88cf1e923a8c9e5bc6953ad412d15a7c70f054d1
galaxyp
parents:
diff changeset
23 from ensembl_rest import get_toplevel, get_transcripts_bed, max_region
da1b538b87e5 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/proteogenomics/retrieve_ensembl_bed commit 88cf1e923a8c9e5bc6953ad412d15a7c70f054d1
galaxyp
parents:
diff changeset
24
da1b538b87e5 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/proteogenomics/retrieve_ensembl_bed commit 88cf1e923a8c9e5bc6953ad412d15a7c70f054d1
galaxyp
parents:
diff changeset
25
da1b538b87e5 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/proteogenomics/retrieve_ensembl_bed commit 88cf1e923a8c9e5bc6953ad412d15a7c70f054d1
galaxyp
parents:
diff changeset
26 def __main__():
da1b538b87e5 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/proteogenomics/retrieve_ensembl_bed commit 88cf1e923a8c9e5bc6953ad412d15a7c70f054d1
galaxyp
parents:
diff changeset
27 parser = argparse.ArgumentParser(
da1b538b87e5 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/proteogenomics/retrieve_ensembl_bed commit 88cf1e923a8c9e5bc6953ad412d15a7c70f054d1
galaxyp
parents:
diff changeset
28 description='Retrieve Ensembl cDNAs in BED format')
da1b538b87e5 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/proteogenomics/retrieve_ensembl_bed commit 88cf1e923a8c9e5bc6953ad412d15a7c70f054d1
galaxyp
parents:
diff changeset
29 parser.add_argument(
da1b538b87e5 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/proteogenomics/retrieve_ensembl_bed commit 88cf1e923a8c9e5bc6953ad412d15a7c70f054d1
galaxyp
parents:
diff changeset
30 '-s', '--species', default='human',
da1b538b87e5 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/proteogenomics/retrieve_ensembl_bed commit 88cf1e923a8c9e5bc6953ad412d15a7c70f054d1
galaxyp
parents:
diff changeset
31 help='Ensembl Species to retrieve')
da1b538b87e5 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/proteogenomics/retrieve_ensembl_bed commit 88cf1e923a8c9e5bc6953ad412d15a7c70f054d1
galaxyp
parents:
diff changeset
32 parser.add_argument(
da1b538b87e5 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/proteogenomics/retrieve_ensembl_bed commit 88cf1e923a8c9e5bc6953ad412d15a7c70f054d1
galaxyp
parents:
diff changeset
33 '-R', '--regions', action='append', default=[],
da1b538b87e5 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/proteogenomics/retrieve_ensembl_bed commit 88cf1e923a8c9e5bc6953ad412d15a7c70f054d1
galaxyp
parents:
diff changeset
34 help='Restrict Ensembl retrieval to regions e.g.:'
da1b538b87e5 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/proteogenomics/retrieve_ensembl_bed commit 88cf1e923a8c9e5bc6953ad412d15a7c70f054d1
galaxyp
parents:
diff changeset
35 + ' X,2:20000-25000,3:100-500+')
da1b538b87e5 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/proteogenomics/retrieve_ensembl_bed commit 88cf1e923a8c9e5bc6953ad412d15a7c70f054d1
galaxyp
parents:
diff changeset
36 parser.add_argument(
1
9c4a48f5d4e7 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/proteogenomics/retrieve_ensembl_bed commit 6babd357845126292cb202aaea0f70ff68819525"
galaxyp
parents: 0
diff changeset
37 '-i', '--interval_file', default=None,
9c4a48f5d4e7 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/proteogenomics/retrieve_ensembl_bed commit 6babd357845126292cb202aaea0f70ff68819525"
galaxyp
parents: 0
diff changeset
38 help='Regions from a bed, gff, or interval file')
9c4a48f5d4e7 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/proteogenomics/retrieve_ensembl_bed commit 6babd357845126292cb202aaea0f70ff68819525"
galaxyp
parents: 0
diff changeset
39 parser.add_argument(
9c4a48f5d4e7 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/proteogenomics/retrieve_ensembl_bed commit 6babd357845126292cb202aaea0f70ff68819525"
galaxyp
parents: 0
diff changeset
40 '-f', '--interval_format', choices=['bed','gff','interval'], default='interval',
9c4a48f5d4e7 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/proteogenomics/retrieve_ensembl_bed commit 6babd357845126292cb202aaea0f70ff68819525"
galaxyp
parents: 0
diff changeset
41 help='Interval format has TAB-separated columns: Seq, Start, End, Strand')
9c4a48f5d4e7 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/proteogenomics/retrieve_ensembl_bed commit 6babd357845126292cb202aaea0f70ff68819525"
galaxyp
parents: 0
diff changeset
42 parser.add_argument(
0
da1b538b87e5 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/proteogenomics/retrieve_ensembl_bed commit 88cf1e923a8c9e5bc6953ad412d15a7c70f054d1
galaxyp
parents:
diff changeset
43 '-B', '--biotypes', action='append', default=[],
da1b538b87e5 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/proteogenomics/retrieve_ensembl_bed commit 88cf1e923a8c9e5bc6953ad412d15a7c70f054d1
galaxyp
parents:
diff changeset
44 help='Restrict Ensembl biotypes to retrieve')
da1b538b87e5 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/proteogenomics/retrieve_ensembl_bed commit 88cf1e923a8c9e5bc6953ad412d15a7c70f054d1
galaxyp
parents:
diff changeset
45 parser.add_argument(
da1b538b87e5 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/proteogenomics/retrieve_ensembl_bed commit 88cf1e923a8c9e5bc6953ad412d15a7c70f054d1
galaxyp
parents:
diff changeset
46 '-X', '--extended_bed', action='store_true', default=False,
da1b538b87e5 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/proteogenomics/retrieve_ensembl_bed commit 88cf1e923a8c9e5bc6953ad412d15a7c70f054d1
galaxyp
parents:
diff changeset
47 help='Include the extended columns returned from Ensembl')
da1b538b87e5 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/proteogenomics/retrieve_ensembl_bed commit 88cf1e923a8c9e5bc6953ad412d15a7c70f054d1
galaxyp
parents:
diff changeset
48 parser.add_argument(
da1b538b87e5 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/proteogenomics/retrieve_ensembl_bed commit 88cf1e923a8c9e5bc6953ad412d15a7c70f054d1
galaxyp
parents:
diff changeset
49 '-U', '--ucsc_chrom_names', action='store_true', default=False,
da1b538b87e5 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/proteogenomics/retrieve_ensembl_bed commit 88cf1e923a8c9e5bc6953ad412d15a7c70f054d1
galaxyp
parents:
diff changeset
50 help='Use the UCSC names for Chromosomes')
da1b538b87e5 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/proteogenomics/retrieve_ensembl_bed commit 88cf1e923a8c9e5bc6953ad412d15a7c70f054d1
galaxyp
parents:
diff changeset
51 parser.add_argument(
da1b538b87e5 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/proteogenomics/retrieve_ensembl_bed commit 88cf1e923a8c9e5bc6953ad412d15a7c70f054d1
galaxyp
parents:
diff changeset
52 '-t', '--toplevel', action='store_true',
da1b538b87e5 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/proteogenomics/retrieve_ensembl_bed commit 88cf1e923a8c9e5bc6953ad412d15a7c70f054d1
galaxyp
parents:
diff changeset
53 help='Print Ensembl toplevel for species')
da1b538b87e5 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/proteogenomics/retrieve_ensembl_bed commit 88cf1e923a8c9e5bc6953ad412d15a7c70f054d1
galaxyp
parents:
diff changeset
54 parser.add_argument(
da1b538b87e5 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/proteogenomics/retrieve_ensembl_bed commit 88cf1e923a8c9e5bc6953ad412d15a7c70f054d1
galaxyp
parents:
diff changeset
55 'output',
da1b538b87e5 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/proteogenomics/retrieve_ensembl_bed commit 88cf1e923a8c9e5bc6953ad412d15a7c70f054d1
galaxyp
parents:
diff changeset
56 help='Output BED filepath, or for stdout: "-"')
da1b538b87e5 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/proteogenomics/retrieve_ensembl_bed commit 88cf1e923a8c9e5bc6953ad412d15a7c70f054d1
galaxyp
parents:
diff changeset
57 parser.add_argument('-v', '--verbose', action='store_true', help='Verbose')
da1b538b87e5 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/proteogenomics/retrieve_ensembl_bed commit 88cf1e923a8c9e5bc6953ad412d15a7c70f054d1
galaxyp
parents:
diff changeset
58 parser.add_argument('-d', '--debug', action='store_true', help='Debug')
da1b538b87e5 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/proteogenomics/retrieve_ensembl_bed commit 88cf1e923a8c9e5bc6953ad412d15a7c70f054d1
galaxyp
parents:
diff changeset
59 args = parser.parse_args()
da1b538b87e5 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/proteogenomics/retrieve_ensembl_bed commit 88cf1e923a8c9e5bc6953ad412d15a7c70f054d1
galaxyp
parents:
diff changeset
60 species = args.species
da1b538b87e5 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/proteogenomics/retrieve_ensembl_bed commit 88cf1e923a8c9e5bc6953ad412d15a7c70f054d1
galaxyp
parents:
diff changeset
61 out_wtr = open(args.output, 'w') if args.output != '-' else sys.stdout
da1b538b87e5 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/proteogenomics/retrieve_ensembl_bed commit 88cf1e923a8c9e5bc6953ad412d15a7c70f054d1
galaxyp
parents:
diff changeset
62 biotypes = ';'.join(['biotype=%s' % bt.strip()
da1b538b87e5 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/proteogenomics/retrieve_ensembl_bed commit 88cf1e923a8c9e5bc6953ad412d15a7c70f054d1
galaxyp
parents:
diff changeset
63 for biotype in args.biotypes
da1b538b87e5 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/proteogenomics/retrieve_ensembl_bed commit 88cf1e923a8c9e5bc6953ad412d15a7c70f054d1
galaxyp
parents:
diff changeset
64 for bt in biotype.split(',') if bt.strip()])
da1b538b87e5 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/proteogenomics/retrieve_ensembl_bed commit 88cf1e923a8c9e5bc6953ad412d15a7c70f054d1
galaxyp
parents:
diff changeset
65
da1b538b87e5 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/proteogenomics/retrieve_ensembl_bed commit 88cf1e923a8c9e5bc6953ad412d15a7c70f054d1
galaxyp
parents:
diff changeset
66 selected_regions = dict() # chrom:(start, end)
da1b538b87e5 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/proteogenomics/retrieve_ensembl_bed commit 88cf1e923a8c9e5bc6953ad412d15a7c70f054d1
galaxyp
parents:
diff changeset
67 region_pat = '^([^:]+)(?::(\d*)(?:-(\d+)([+-])?)?)?'
da1b538b87e5 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/proteogenomics/retrieve_ensembl_bed commit 88cf1e923a8c9e5bc6953ad412d15a7c70f054d1
galaxyp
parents:
diff changeset
68 if args.regions:
da1b538b87e5 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/proteogenomics/retrieve_ensembl_bed commit 88cf1e923a8c9e5bc6953ad412d15a7c70f054d1
galaxyp
parents:
diff changeset
69 for entry in args.regions:
da1b538b87e5 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/proteogenomics/retrieve_ensembl_bed commit 88cf1e923a8c9e5bc6953ad412d15a7c70f054d1
galaxyp
parents:
diff changeset
70 if not entry:
da1b538b87e5 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/proteogenomics/retrieve_ensembl_bed commit 88cf1e923a8c9e5bc6953ad412d15a7c70f054d1
galaxyp
parents:
diff changeset
71 continue
da1b538b87e5 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/proteogenomics/retrieve_ensembl_bed commit 88cf1e923a8c9e5bc6953ad412d15a7c70f054d1
galaxyp
parents:
diff changeset
72 regs = [x.strip() for x in entry.split(',') if x.strip()]
da1b538b87e5 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/proteogenomics/retrieve_ensembl_bed commit 88cf1e923a8c9e5bc6953ad412d15a7c70f054d1
galaxyp
parents:
diff changeset
73 for reg in regs:
da1b538b87e5 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/proteogenomics/retrieve_ensembl_bed commit 88cf1e923a8c9e5bc6953ad412d15a7c70f054d1
galaxyp
parents:
diff changeset
74 m = re.match(region_pat, reg)
da1b538b87e5 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/proteogenomics/retrieve_ensembl_bed commit 88cf1e923a8c9e5bc6953ad412d15a7c70f054d1
galaxyp
parents:
diff changeset
75 if m:
da1b538b87e5 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/proteogenomics/retrieve_ensembl_bed commit 88cf1e923a8c9e5bc6953ad412d15a7c70f054d1
galaxyp
parents:
diff changeset
76 (chrom, start, end, strand) = m.groups()
da1b538b87e5 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/proteogenomics/retrieve_ensembl_bed commit 88cf1e923a8c9e5bc6953ad412d15a7c70f054d1
galaxyp
parents:
diff changeset
77 if chrom:
da1b538b87e5 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/proteogenomics/retrieve_ensembl_bed commit 88cf1e923a8c9e5bc6953ad412d15a7c70f054d1
galaxyp
parents:
diff changeset
78 if chrom not in selected_regions:
da1b538b87e5 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/proteogenomics/retrieve_ensembl_bed commit 88cf1e923a8c9e5bc6953ad412d15a7c70f054d1
galaxyp
parents:
diff changeset
79 selected_regions[chrom] = []
da1b538b87e5 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/proteogenomics/retrieve_ensembl_bed commit 88cf1e923a8c9e5bc6953ad412d15a7c70f054d1
galaxyp
parents:
diff changeset
80 selected_regions[chrom].append([start, end, strand])
da1b538b87e5 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/proteogenomics/retrieve_ensembl_bed commit 88cf1e923a8c9e5bc6953ad412d15a7c70f054d1
galaxyp
parents:
diff changeset
81 if args.debug:
da1b538b87e5 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/proteogenomics/retrieve_ensembl_bed commit 88cf1e923a8c9e5bc6953ad412d15a7c70f054d1
galaxyp
parents:
diff changeset
82 print("selected_regions: %s" % selected_regions, file=sys.stderr)
da1b538b87e5 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/proteogenomics/retrieve_ensembl_bed commit 88cf1e923a8c9e5bc6953ad412d15a7c70f054d1
galaxyp
parents:
diff changeset
83
1
9c4a48f5d4e7 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/proteogenomics/retrieve_ensembl_bed commit 6babd357845126292cb202aaea0f70ff68819525"
galaxyp
parents: 0
diff changeset
84 if args.interval_file:
9c4a48f5d4e7 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/proteogenomics/retrieve_ensembl_bed commit 6babd357845126292cb202aaea0f70ff68819525"
galaxyp
parents: 0
diff changeset
85 pat = r'^(?:chr)?([^\t]+)(?:\t(\d+)(?:\t(\d+)(?:\t([+-])?)?)?)?.*'
9c4a48f5d4e7 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/proteogenomics/retrieve_ensembl_bed commit 6babd357845126292cb202aaea0f70ff68819525"
galaxyp
parents: 0
diff changeset
86 if args.interval_format == 'bed':
9c4a48f5d4e7 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/proteogenomics/retrieve_ensembl_bed commit 6babd357845126292cb202aaea0f70ff68819525"
galaxyp
parents: 0
diff changeset
87 pat = r'^(?:chr)?([^\t]+)\t(\d+)\t(\d+)(?:(?:\t[^\t]+\t[^\t]+\t)([+-]))?.*'
9c4a48f5d4e7 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/proteogenomics/retrieve_ensembl_bed commit 6babd357845126292cb202aaea0f70ff68819525"
galaxyp
parents: 0
diff changeset
88 elif args.interval_format == 'gff':
9c4a48f5d4e7 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/proteogenomics/retrieve_ensembl_bed commit 6babd357845126292cb202aaea0f70ff68819525"
galaxyp
parents: 0
diff changeset
89 pat = r'^(?:chr)?([^\t]+)\t(\d+)\t(\d+)(?:(?:\t[^\t]+\t[^\t]+\t)([+-]))?.*'
9c4a48f5d4e7 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/proteogenomics/retrieve_ensembl_bed commit 6babd357845126292cb202aaea0f70ff68819525"
galaxyp
parents: 0
diff changeset
90 with open(args.interval_file,'r') as fh:
9c4a48f5d4e7 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/proteogenomics/retrieve_ensembl_bed commit 6babd357845126292cb202aaea0f70ff68819525"
galaxyp
parents: 0
diff changeset
91 for i, line in enumerate(fh):
9c4a48f5d4e7 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/proteogenomics/retrieve_ensembl_bed commit 6babd357845126292cb202aaea0f70ff68819525"
galaxyp
parents: 0
diff changeset
92 if line.startswith('#'):
9c4a48f5d4e7 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/proteogenomics/retrieve_ensembl_bed commit 6babd357845126292cb202aaea0f70ff68819525"
galaxyp
parents: 0
diff changeset
93 continue
9c4a48f5d4e7 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/proteogenomics/retrieve_ensembl_bed commit 6babd357845126292cb202aaea0f70ff68819525"
galaxyp
parents: 0
diff changeset
94 m = re.match(pat, line.rstrip())
9c4a48f5d4e7 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/proteogenomics/retrieve_ensembl_bed commit 6babd357845126292cb202aaea0f70ff68819525"
galaxyp
parents: 0
diff changeset
95 if m:
9c4a48f5d4e7 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/proteogenomics/retrieve_ensembl_bed commit 6babd357845126292cb202aaea0f70ff68819525"
galaxyp
parents: 0
diff changeset
96 (chrom, start, end, strand) = m.groups()
9c4a48f5d4e7 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/proteogenomics/retrieve_ensembl_bed commit 6babd357845126292cb202aaea0f70ff68819525"
galaxyp
parents: 0
diff changeset
97 if chrom:
9c4a48f5d4e7 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/proteogenomics/retrieve_ensembl_bed commit 6babd357845126292cb202aaea0f70ff68819525"
galaxyp
parents: 0
diff changeset
98 if chrom not in selected_regions:
9c4a48f5d4e7 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/proteogenomics/retrieve_ensembl_bed commit 6babd357845126292cb202aaea0f70ff68819525"
galaxyp
parents: 0
diff changeset
99 selected_regions[chrom] = []
9c4a48f5d4e7 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/proteogenomics/retrieve_ensembl_bed commit 6babd357845126292cb202aaea0f70ff68819525"
galaxyp
parents: 0
diff changeset
100 selected_regions[chrom].append([start, end, strand])
9c4a48f5d4e7 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/proteogenomics/retrieve_ensembl_bed commit 6babd357845126292cb202aaea0f70ff68819525"
galaxyp
parents: 0
diff changeset
101 if args.debug:
9c4a48f5d4e7 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/proteogenomics/retrieve_ensembl_bed commit 6babd357845126292cb202aaea0f70ff68819525"
galaxyp
parents: 0
diff changeset
102 print("selected_regions: %s" % selected_regions, file=sys.stderr)
9c4a48f5d4e7 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/proteogenomics/retrieve_ensembl_bed commit 6babd357845126292cb202aaea0f70ff68819525"
galaxyp
parents: 0
diff changeset
103
9c4a48f5d4e7 "planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/proteogenomics/retrieve_ensembl_bed commit 6babd357845126292cb202aaea0f70ff68819525"
galaxyp
parents: 0
diff changeset
104
0
da1b538b87e5 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/proteogenomics/retrieve_ensembl_bed commit 88cf1e923a8c9e5bc6953ad412d15a7c70f054d1
galaxyp
parents:
diff changeset
105 def retrieve_region(species, ref, start, stop, strand):
da1b538b87e5 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/proteogenomics/retrieve_ensembl_bed commit 88cf1e923a8c9e5bc6953ad412d15a7c70f054d1
galaxyp
parents:
diff changeset
106 transcript_count = 0
da1b538b87e5 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/proteogenomics/retrieve_ensembl_bed commit 88cf1e923a8c9e5bc6953ad412d15a7c70f054d1
galaxyp
parents:
diff changeset
107 regions = list(range(start, stop, max_region))
da1b538b87e5 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/proteogenomics/retrieve_ensembl_bed commit 88cf1e923a8c9e5bc6953ad412d15a7c70f054d1
galaxyp
parents:
diff changeset
108 if not regions or regions[-1] < stop:
da1b538b87e5 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/proteogenomics/retrieve_ensembl_bed commit 88cf1e923a8c9e5bc6953ad412d15a7c70f054d1
galaxyp
parents:
diff changeset
109 regions.append(stop)
da1b538b87e5 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/proteogenomics/retrieve_ensembl_bed commit 88cf1e923a8c9e5bc6953ad412d15a7c70f054d1
galaxyp
parents:
diff changeset
110 for end in regions[1:]:
da1b538b87e5 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/proteogenomics/retrieve_ensembl_bed commit 88cf1e923a8c9e5bc6953ad412d15a7c70f054d1
galaxyp
parents:
diff changeset
111 bedlines = get_transcripts_bed(species, ref, start, end,
da1b538b87e5 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/proteogenomics/retrieve_ensembl_bed commit 88cf1e923a8c9e5bc6953ad412d15a7c70f054d1
galaxyp
parents:
diff changeset
112 strand=strand, params=biotypes)
da1b538b87e5 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/proteogenomics/retrieve_ensembl_bed commit 88cf1e923a8c9e5bc6953ad412d15a7c70f054d1
galaxyp
parents:
diff changeset
113 if args.debug:
da1b538b87e5 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/proteogenomics/retrieve_ensembl_bed commit 88cf1e923a8c9e5bc6953ad412d15a7c70f054d1
galaxyp
parents:
diff changeset
114 print("%s\t%s\tstart: %d\tend: %d\tcDNA transcripts:%d" %
da1b538b87e5 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/proteogenomics/retrieve_ensembl_bed commit 88cf1e923a8c9e5bc6953ad412d15a7c70f054d1
galaxyp
parents:
diff changeset
115 (species, ref, start, end, len(bedlines)),
da1b538b87e5 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/proteogenomics/retrieve_ensembl_bed commit 88cf1e923a8c9e5bc6953ad412d15a7c70f054d1
galaxyp
parents:
diff changeset
116 file=sys.stderr)
da1b538b87e5 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/proteogenomics/retrieve_ensembl_bed commit 88cf1e923a8c9e5bc6953ad412d15a7c70f054d1
galaxyp
parents:
diff changeset
117 # start, end, seq
da1b538b87e5 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/proteogenomics/retrieve_ensembl_bed commit 88cf1e923a8c9e5bc6953ad412d15a7c70f054d1
galaxyp
parents:
diff changeset
118 for i, bedline in enumerate(bedlines):
da1b538b87e5 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/proteogenomics/retrieve_ensembl_bed commit 88cf1e923a8c9e5bc6953ad412d15a7c70f054d1
galaxyp
parents:
diff changeset
119 if args.debug:
da1b538b87e5 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/proteogenomics/retrieve_ensembl_bed commit 88cf1e923a8c9e5bc6953ad412d15a7c70f054d1
galaxyp
parents:
diff changeset
120 print("%s\n" % (bedline), file=sys.stderr)
da1b538b87e5 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/proteogenomics/retrieve_ensembl_bed commit 88cf1e923a8c9e5bc6953ad412d15a7c70f054d1
galaxyp
parents:
diff changeset
121 if not args.ucsc_chrom_names:
da1b538b87e5 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/proteogenomics/retrieve_ensembl_bed commit 88cf1e923a8c9e5bc6953ad412d15a7c70f054d1
galaxyp
parents:
diff changeset
122 bedline = re.sub('^[^\t]+', ref, bedline)
da1b538b87e5 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/proteogenomics/retrieve_ensembl_bed commit 88cf1e923a8c9e5bc6953ad412d15a7c70f054d1
galaxyp
parents:
diff changeset
123 try:
da1b538b87e5 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/proteogenomics/retrieve_ensembl_bed commit 88cf1e923a8c9e5bc6953ad412d15a7c70f054d1
galaxyp
parents:
diff changeset
124 if out_wtr:
da1b538b87e5 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/proteogenomics/retrieve_ensembl_bed commit 88cf1e923a8c9e5bc6953ad412d15a7c70f054d1
galaxyp
parents:
diff changeset
125 out_wtr.write(bedline.replace(',\t', '\t')
da1b538b87e5 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/proteogenomics/retrieve_ensembl_bed commit 88cf1e923a8c9e5bc6953ad412d15a7c70f054d1
galaxyp
parents:
diff changeset
126 if args.extended_bed
da1b538b87e5 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/proteogenomics/retrieve_ensembl_bed commit 88cf1e923a8c9e5bc6953ad412d15a7c70f054d1
galaxyp
parents:
diff changeset
127 else str(bed_from_line(bedline)))
da1b538b87e5 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/proteogenomics/retrieve_ensembl_bed commit 88cf1e923a8c9e5bc6953ad412d15a7c70f054d1
galaxyp
parents:
diff changeset
128 out_wtr.write("\n")
da1b538b87e5 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/proteogenomics/retrieve_ensembl_bed commit 88cf1e923a8c9e5bc6953ad412d15a7c70f054d1
galaxyp
parents:
diff changeset
129 out_wtr.flush()
da1b538b87e5 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/proteogenomics/retrieve_ensembl_bed commit 88cf1e923a8c9e5bc6953ad412d15a7c70f054d1
galaxyp
parents:
diff changeset
130 except Exception as e:
da1b538b87e5 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/proteogenomics/retrieve_ensembl_bed commit 88cf1e923a8c9e5bc6953ad412d15a7c70f054d1
galaxyp
parents:
diff changeset
131 print("BED error (%s) : %s\n" % (e, bedline),
da1b538b87e5 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/proteogenomics/retrieve_ensembl_bed commit 88cf1e923a8c9e5bc6953ad412d15a7c70f054d1
galaxyp
parents:
diff changeset
132 file=sys.stderr)
da1b538b87e5 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/proteogenomics/retrieve_ensembl_bed commit 88cf1e923a8c9e5bc6953ad412d15a7c70f054d1
galaxyp
parents:
diff changeset
133 start = end + 1
da1b538b87e5 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/proteogenomics/retrieve_ensembl_bed commit 88cf1e923a8c9e5bc6953ad412d15a7c70f054d1
galaxyp
parents:
diff changeset
134 return transcript_count
da1b538b87e5 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/proteogenomics/retrieve_ensembl_bed commit 88cf1e923a8c9e5bc6953ad412d15a7c70f054d1
galaxyp
parents:
diff changeset
135
da1b538b87e5 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/proteogenomics/retrieve_ensembl_bed commit 88cf1e923a8c9e5bc6953ad412d15a7c70f054d1
galaxyp
parents:
diff changeset
136 coord_systems = get_toplevel(species)
da1b538b87e5 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/proteogenomics/retrieve_ensembl_bed commit 88cf1e923a8c9e5bc6953ad412d15a7c70f054d1
galaxyp
parents:
diff changeset
137 if 'chromosome' in coord_systems:
da1b538b87e5 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/proteogenomics/retrieve_ensembl_bed commit 88cf1e923a8c9e5bc6953ad412d15a7c70f054d1
galaxyp
parents:
diff changeset
138 ref_lengths = dict()
da1b538b87e5 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/proteogenomics/retrieve_ensembl_bed commit 88cf1e923a8c9e5bc6953ad412d15a7c70f054d1
galaxyp
parents:
diff changeset
139 for ref in sorted(coord_systems['chromosome'].keys()):
da1b538b87e5 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/proteogenomics/retrieve_ensembl_bed commit 88cf1e923a8c9e5bc6953ad412d15a7c70f054d1
galaxyp
parents:
diff changeset
140 length = coord_systems['chromosome'][ref]
da1b538b87e5 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/proteogenomics/retrieve_ensembl_bed commit 88cf1e923a8c9e5bc6953ad412d15a7c70f054d1
galaxyp
parents:
diff changeset
141 ref_lengths[ref] = length
da1b538b87e5 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/proteogenomics/retrieve_ensembl_bed commit 88cf1e923a8c9e5bc6953ad412d15a7c70f054d1
galaxyp
parents:
diff changeset
142 if args.toplevel:
da1b538b87e5 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/proteogenomics/retrieve_ensembl_bed commit 88cf1e923a8c9e5bc6953ad412d15a7c70f054d1
galaxyp
parents:
diff changeset
143 print("%s\t%s\tlength: %d" % (species, ref, length),
da1b538b87e5 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/proteogenomics/retrieve_ensembl_bed commit 88cf1e923a8c9e5bc6953ad412d15a7c70f054d1
galaxyp
parents:
diff changeset
144 file=sys.stderr)
da1b538b87e5 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/proteogenomics/retrieve_ensembl_bed commit 88cf1e923a8c9e5bc6953ad412d15a7c70f054d1
galaxyp
parents:
diff changeset
145 if selected_regions:
da1b538b87e5 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/proteogenomics/retrieve_ensembl_bed commit 88cf1e923a8c9e5bc6953ad412d15a7c70f054d1
galaxyp
parents:
diff changeset
146 transcript_count = 0
da1b538b87e5 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/proteogenomics/retrieve_ensembl_bed commit 88cf1e923a8c9e5bc6953ad412d15a7c70f054d1
galaxyp
parents:
diff changeset
147 for ref in sorted(selected_regions.keys()):
da1b538b87e5 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/proteogenomics/retrieve_ensembl_bed commit 88cf1e923a8c9e5bc6953ad412d15a7c70f054d1
galaxyp
parents:
diff changeset
148 if ref in ref_lengths:
da1b538b87e5 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/proteogenomics/retrieve_ensembl_bed commit 88cf1e923a8c9e5bc6953ad412d15a7c70f054d1
galaxyp
parents:
diff changeset
149 for reg in selected_regions[ref]:
da1b538b87e5 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/proteogenomics/retrieve_ensembl_bed commit 88cf1e923a8c9e5bc6953ad412d15a7c70f054d1
galaxyp
parents:
diff changeset
150 (_start, _stop, _strand) = reg
da1b538b87e5 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/proteogenomics/retrieve_ensembl_bed commit 88cf1e923a8c9e5bc6953ad412d15a7c70f054d1
galaxyp
parents:
diff changeset
151 start = int(_start) if _start else 0
da1b538b87e5 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/proteogenomics/retrieve_ensembl_bed commit 88cf1e923a8c9e5bc6953ad412d15a7c70f054d1
galaxyp
parents:
diff changeset
152 stop = int(_stop) if _stop else ref_lengths[ref]
da1b538b87e5 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/proteogenomics/retrieve_ensembl_bed commit 88cf1e923a8c9e5bc6953ad412d15a7c70f054d1
galaxyp
parents:
diff changeset
153 strand = '' if not _strand else ':1'\
da1b538b87e5 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/proteogenomics/retrieve_ensembl_bed commit 88cf1e923a8c9e5bc6953ad412d15a7c70f054d1
galaxyp
parents:
diff changeset
154 if _strand == '+' else ':-1'
da1b538b87e5 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/proteogenomics/retrieve_ensembl_bed commit 88cf1e923a8c9e5bc6953ad412d15a7c70f054d1
galaxyp
parents:
diff changeset
155 transcript_count += retrieve_region(species, ref,
da1b538b87e5 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/proteogenomics/retrieve_ensembl_bed commit 88cf1e923a8c9e5bc6953ad412d15a7c70f054d1
galaxyp
parents:
diff changeset
156 start, stop,
da1b538b87e5 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/proteogenomics/retrieve_ensembl_bed commit 88cf1e923a8c9e5bc6953ad412d15a7c70f054d1
galaxyp
parents:
diff changeset
157 strand)
da1b538b87e5 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/proteogenomics/retrieve_ensembl_bed commit 88cf1e923a8c9e5bc6953ad412d15a7c70f054d1
galaxyp
parents:
diff changeset
158 if args.debug or args.verbose:
da1b538b87e5 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/proteogenomics/retrieve_ensembl_bed commit 88cf1e923a8c9e5bc6953ad412d15a7c70f054d1
galaxyp
parents:
diff changeset
159 length = stop - start
da1b538b87e5 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/proteogenomics/retrieve_ensembl_bed commit 88cf1e923a8c9e5bc6953ad412d15a7c70f054d1
galaxyp
parents:
diff changeset
160 print("%s\t%s:%d-%d%s\tlength: %d\ttrancripts:%d" %
da1b538b87e5 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/proteogenomics/retrieve_ensembl_bed commit 88cf1e923a8c9e5bc6953ad412d15a7c70f054d1
galaxyp
parents:
diff changeset
161 (species, ref, start, stop, strand,
da1b538b87e5 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/proteogenomics/retrieve_ensembl_bed commit 88cf1e923a8c9e5bc6953ad412d15a7c70f054d1
galaxyp
parents:
diff changeset
162 length, transcript_count),
da1b538b87e5 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/proteogenomics/retrieve_ensembl_bed commit 88cf1e923a8c9e5bc6953ad412d15a7c70f054d1
galaxyp
parents:
diff changeset
163 file=sys.stderr)
da1b538b87e5 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/proteogenomics/retrieve_ensembl_bed commit 88cf1e923a8c9e5bc6953ad412d15a7c70f054d1
galaxyp
parents:
diff changeset
164 else:
da1b538b87e5 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/proteogenomics/retrieve_ensembl_bed commit 88cf1e923a8c9e5bc6953ad412d15a7c70f054d1
galaxyp
parents:
diff changeset
165 strand = ''
da1b538b87e5 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/proteogenomics/retrieve_ensembl_bed commit 88cf1e923a8c9e5bc6953ad412d15a7c70f054d1
galaxyp
parents:
diff changeset
166 start = 0
da1b538b87e5 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/proteogenomics/retrieve_ensembl_bed commit 88cf1e923a8c9e5bc6953ad412d15a7c70f054d1
galaxyp
parents:
diff changeset
167 for ref in sorted(ref_lengths.keys()):
da1b538b87e5 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/proteogenomics/retrieve_ensembl_bed commit 88cf1e923a8c9e5bc6953ad412d15a7c70f054d1
galaxyp
parents:
diff changeset
168 length = ref_lengths[ref]
da1b538b87e5 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/proteogenomics/retrieve_ensembl_bed commit 88cf1e923a8c9e5bc6953ad412d15a7c70f054d1
galaxyp
parents:
diff changeset
169 transcript_count = 0
da1b538b87e5 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/proteogenomics/retrieve_ensembl_bed commit 88cf1e923a8c9e5bc6953ad412d15a7c70f054d1
galaxyp
parents:
diff changeset
170 if args.debug:
da1b538b87e5 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/proteogenomics/retrieve_ensembl_bed commit 88cf1e923a8c9e5bc6953ad412d15a7c70f054d1
galaxyp
parents:
diff changeset
171 print("Retrieving transcripts: %s\t%s\tlength: %d" %
da1b538b87e5 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/proteogenomics/retrieve_ensembl_bed commit 88cf1e923a8c9e5bc6953ad412d15a7c70f054d1
galaxyp
parents:
diff changeset
172 (species, ref, length), file=sys.stderr)
da1b538b87e5 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/proteogenomics/retrieve_ensembl_bed commit 88cf1e923a8c9e5bc6953ad412d15a7c70f054d1
galaxyp
parents:
diff changeset
173 transcript_count += retrieve_region(species, ref, start,
da1b538b87e5 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/proteogenomics/retrieve_ensembl_bed commit 88cf1e923a8c9e5bc6953ad412d15a7c70f054d1
galaxyp
parents:
diff changeset
174 length, strand)
da1b538b87e5 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/proteogenomics/retrieve_ensembl_bed commit 88cf1e923a8c9e5bc6953ad412d15a7c70f054d1
galaxyp
parents:
diff changeset
175 if args.debug or args.verbose:
da1b538b87e5 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/proteogenomics/retrieve_ensembl_bed commit 88cf1e923a8c9e5bc6953ad412d15a7c70f054d1
galaxyp
parents:
diff changeset
176 print("%s\t%s\tlength: %d\ttrancripts:%d" %
da1b538b87e5 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/proteogenomics/retrieve_ensembl_bed commit 88cf1e923a8c9e5bc6953ad412d15a7c70f054d1
galaxyp
parents:
diff changeset
177 (species, ref, length, transcript_count),
da1b538b87e5 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/proteogenomics/retrieve_ensembl_bed commit 88cf1e923a8c9e5bc6953ad412d15a7c70f054d1
galaxyp
parents:
diff changeset
178 file=sys.stderr)
da1b538b87e5 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/proteogenomics/retrieve_ensembl_bed commit 88cf1e923a8c9e5bc6953ad412d15a7c70f054d1
galaxyp
parents:
diff changeset
179
da1b538b87e5 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/proteogenomics/retrieve_ensembl_bed commit 88cf1e923a8c9e5bc6953ad412d15a7c70f054d1
galaxyp
parents:
diff changeset
180
da1b538b87e5 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/proteogenomics/retrieve_ensembl_bed commit 88cf1e923a8c9e5bc6953ad412d15a7c70f054d1
galaxyp
parents:
diff changeset
181 if __name__ == "__main__":
da1b538b87e5 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/proteogenomics/retrieve_ensembl_bed commit 88cf1e923a8c9e5bc6953ad412d15a7c70f054d1
galaxyp
parents:
diff changeset
182 __main__()