annotate retrieve_ensembl_bed.py @ 0:887e111c0919 draft

planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/proteogenomics/retrieve_ensembl_bed commit 3fd7be931712e7fa5b281bc8c48104c8583ef7f0
author galaxyp
date Sun, 14 Jan 2018 14:11:53 -0500
parents
children c3d600729b6f
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
rev   line source
0
887e111c0919 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/proteogenomics/retrieve_ensembl_bed commit 3fd7be931712e7fa5b281bc8c48104c8583ef7f0
galaxyp
parents:
diff changeset
1 #!/usr/bin/env python
887e111c0919 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/proteogenomics/retrieve_ensembl_bed commit 3fd7be931712e7fa5b281bc8c48104c8583ef7f0
galaxyp
parents:
diff changeset
2 """
887e111c0919 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/proteogenomics/retrieve_ensembl_bed commit 3fd7be931712e7fa5b281bc8c48104c8583ef7f0
galaxyp
parents:
diff changeset
3 #
887e111c0919 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/proteogenomics/retrieve_ensembl_bed commit 3fd7be931712e7fa5b281bc8c48104c8583ef7f0
galaxyp
parents:
diff changeset
4 #------------------------------------------------------------------------------
887e111c0919 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/proteogenomics/retrieve_ensembl_bed commit 3fd7be931712e7fa5b281bc8c48104c8583ef7f0
galaxyp
parents:
diff changeset
5 # University of Minnesota
887e111c0919 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/proteogenomics/retrieve_ensembl_bed commit 3fd7be931712e7fa5b281bc8c48104c8583ef7f0
galaxyp
parents:
diff changeset
6 # Copyright 2017, Regents of the University of Minnesota
887e111c0919 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/proteogenomics/retrieve_ensembl_bed commit 3fd7be931712e7fa5b281bc8c48104c8583ef7f0
galaxyp
parents:
diff changeset
7 #------------------------------------------------------------------------------
887e111c0919 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/proteogenomics/retrieve_ensembl_bed commit 3fd7be931712e7fa5b281bc8c48104c8583ef7f0
galaxyp
parents:
diff changeset
8 # Author:
887e111c0919 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/proteogenomics/retrieve_ensembl_bed commit 3fd7be931712e7fa5b281bc8c48104c8583ef7f0
galaxyp
parents:
diff changeset
9 #
887e111c0919 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/proteogenomics/retrieve_ensembl_bed commit 3fd7be931712e7fa5b281bc8c48104c8583ef7f0
galaxyp
parents:
diff changeset
10 # James E Johnson
887e111c0919 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/proteogenomics/retrieve_ensembl_bed commit 3fd7be931712e7fa5b281bc8c48104c8583ef7f0
galaxyp
parents:
diff changeset
11 #
887e111c0919 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/proteogenomics/retrieve_ensembl_bed commit 3fd7be931712e7fa5b281bc8c48104c8583ef7f0
galaxyp
parents:
diff changeset
12 #------------------------------------------------------------------------------
887e111c0919 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/proteogenomics/retrieve_ensembl_bed commit 3fd7be931712e7fa5b281bc8c48104c8583ef7f0
galaxyp
parents:
diff changeset
13 """
887e111c0919 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/proteogenomics/retrieve_ensembl_bed commit 3fd7be931712e7fa5b281bc8c48104c8583ef7f0
galaxyp
parents:
diff changeset
14
887e111c0919 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/proteogenomics/retrieve_ensembl_bed commit 3fd7be931712e7fa5b281bc8c48104c8583ef7f0
galaxyp
parents:
diff changeset
15 import argparse
887e111c0919 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/proteogenomics/retrieve_ensembl_bed commit 3fd7be931712e7fa5b281bc8c48104c8583ef7f0
galaxyp
parents:
diff changeset
16 import re
887e111c0919 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/proteogenomics/retrieve_ensembl_bed commit 3fd7be931712e7fa5b281bc8c48104c8583ef7f0
galaxyp
parents:
diff changeset
17 import sys
887e111c0919 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/proteogenomics/retrieve_ensembl_bed commit 3fd7be931712e7fa5b281bc8c48104c8583ef7f0
galaxyp
parents:
diff changeset
18
887e111c0919 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/proteogenomics/retrieve_ensembl_bed commit 3fd7be931712e7fa5b281bc8c48104c8583ef7f0
galaxyp
parents:
diff changeset
19 from bedutil import bed_from_line
887e111c0919 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/proteogenomics/retrieve_ensembl_bed commit 3fd7be931712e7fa5b281bc8c48104c8583ef7f0
galaxyp
parents:
diff changeset
20
887e111c0919 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/proteogenomics/retrieve_ensembl_bed commit 3fd7be931712e7fa5b281bc8c48104c8583ef7f0
galaxyp
parents:
diff changeset
21 from ensembl_rest import get_toplevel, get_transcripts_bed, max_region
887e111c0919 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/proteogenomics/retrieve_ensembl_bed commit 3fd7be931712e7fa5b281bc8c48104c8583ef7f0
galaxyp
parents:
diff changeset
22
887e111c0919 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/proteogenomics/retrieve_ensembl_bed commit 3fd7be931712e7fa5b281bc8c48104c8583ef7f0
galaxyp
parents:
diff changeset
23
887e111c0919 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/proteogenomics/retrieve_ensembl_bed commit 3fd7be931712e7fa5b281bc8c48104c8583ef7f0
galaxyp
parents:
diff changeset
24 def __main__():
887e111c0919 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/proteogenomics/retrieve_ensembl_bed commit 3fd7be931712e7fa5b281bc8c48104c8583ef7f0
galaxyp
parents:
diff changeset
25 parser = argparse.ArgumentParser(
887e111c0919 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/proteogenomics/retrieve_ensembl_bed commit 3fd7be931712e7fa5b281bc8c48104c8583ef7f0
galaxyp
parents:
diff changeset
26 description='Retrieve Ensembl cDNAs in BED format')
887e111c0919 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/proteogenomics/retrieve_ensembl_bed commit 3fd7be931712e7fa5b281bc8c48104c8583ef7f0
galaxyp
parents:
diff changeset
27 parser.add_argument(
887e111c0919 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/proteogenomics/retrieve_ensembl_bed commit 3fd7be931712e7fa5b281bc8c48104c8583ef7f0
galaxyp
parents:
diff changeset
28 '-s', '--species', default='human',
887e111c0919 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/proteogenomics/retrieve_ensembl_bed commit 3fd7be931712e7fa5b281bc8c48104c8583ef7f0
galaxyp
parents:
diff changeset
29 help='Ensembl Species to retrieve')
887e111c0919 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/proteogenomics/retrieve_ensembl_bed commit 3fd7be931712e7fa5b281bc8c48104c8583ef7f0
galaxyp
parents:
diff changeset
30 parser.add_argument(
887e111c0919 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/proteogenomics/retrieve_ensembl_bed commit 3fd7be931712e7fa5b281bc8c48104c8583ef7f0
galaxyp
parents:
diff changeset
31 '-R', '--regions', action='append', default=[],
887e111c0919 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/proteogenomics/retrieve_ensembl_bed commit 3fd7be931712e7fa5b281bc8c48104c8583ef7f0
galaxyp
parents:
diff changeset
32 help='Restrict Ensembl retrieval to regions e.g.:'
887e111c0919 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/proteogenomics/retrieve_ensembl_bed commit 3fd7be931712e7fa5b281bc8c48104c8583ef7f0
galaxyp
parents:
diff changeset
33 + ' X,2:20000-25000,3:100-500+')
887e111c0919 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/proteogenomics/retrieve_ensembl_bed commit 3fd7be931712e7fa5b281bc8c48104c8583ef7f0
galaxyp
parents:
diff changeset
34 parser.add_argument(
887e111c0919 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/proteogenomics/retrieve_ensembl_bed commit 3fd7be931712e7fa5b281bc8c48104c8583ef7f0
galaxyp
parents:
diff changeset
35 '-B', '--biotypes', action='append', default=[],
887e111c0919 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/proteogenomics/retrieve_ensembl_bed commit 3fd7be931712e7fa5b281bc8c48104c8583ef7f0
galaxyp
parents:
diff changeset
36 help='Restrict Ensembl biotypes to retrieve')
887e111c0919 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/proteogenomics/retrieve_ensembl_bed commit 3fd7be931712e7fa5b281bc8c48104c8583ef7f0
galaxyp
parents:
diff changeset
37 parser.add_argument(
887e111c0919 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/proteogenomics/retrieve_ensembl_bed commit 3fd7be931712e7fa5b281bc8c48104c8583ef7f0
galaxyp
parents:
diff changeset
38 '-X', '--extended_bed', action='store_true', default=False,
887e111c0919 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/proteogenomics/retrieve_ensembl_bed commit 3fd7be931712e7fa5b281bc8c48104c8583ef7f0
galaxyp
parents:
diff changeset
39 help='Include the extended columns returned from Ensembl')
887e111c0919 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/proteogenomics/retrieve_ensembl_bed commit 3fd7be931712e7fa5b281bc8c48104c8583ef7f0
galaxyp
parents:
diff changeset
40 parser.add_argument(
887e111c0919 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/proteogenomics/retrieve_ensembl_bed commit 3fd7be931712e7fa5b281bc8c48104c8583ef7f0
galaxyp
parents:
diff changeset
41 '-U', '--ucsc_chrom_names', action='store_true', default=False,
887e111c0919 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/proteogenomics/retrieve_ensembl_bed commit 3fd7be931712e7fa5b281bc8c48104c8583ef7f0
galaxyp
parents:
diff changeset
42 help='Use the UCSC names for Chromosomes')
887e111c0919 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/proteogenomics/retrieve_ensembl_bed commit 3fd7be931712e7fa5b281bc8c48104c8583ef7f0
galaxyp
parents:
diff changeset
43 parser.add_argument(
887e111c0919 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/proteogenomics/retrieve_ensembl_bed commit 3fd7be931712e7fa5b281bc8c48104c8583ef7f0
galaxyp
parents:
diff changeset
44 '-t', '--toplevel', action='store_true',
887e111c0919 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/proteogenomics/retrieve_ensembl_bed commit 3fd7be931712e7fa5b281bc8c48104c8583ef7f0
galaxyp
parents:
diff changeset
45 help='Print Ensembl toplevel for species')
887e111c0919 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/proteogenomics/retrieve_ensembl_bed commit 3fd7be931712e7fa5b281bc8c48104c8583ef7f0
galaxyp
parents:
diff changeset
46 parser.add_argument(
887e111c0919 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/proteogenomics/retrieve_ensembl_bed commit 3fd7be931712e7fa5b281bc8c48104c8583ef7f0
galaxyp
parents:
diff changeset
47 'output',
887e111c0919 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/proteogenomics/retrieve_ensembl_bed commit 3fd7be931712e7fa5b281bc8c48104c8583ef7f0
galaxyp
parents:
diff changeset
48 help='Output BED filepath, or for stdout: "-"')
887e111c0919 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/proteogenomics/retrieve_ensembl_bed commit 3fd7be931712e7fa5b281bc8c48104c8583ef7f0
galaxyp
parents:
diff changeset
49 parser.add_argument('-v', '--verbose', action='store_true', help='Verbose')
887e111c0919 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/proteogenomics/retrieve_ensembl_bed commit 3fd7be931712e7fa5b281bc8c48104c8583ef7f0
galaxyp
parents:
diff changeset
50 parser.add_argument('-d', '--debug', action='store_true', help='Debug')
887e111c0919 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/proteogenomics/retrieve_ensembl_bed commit 3fd7be931712e7fa5b281bc8c48104c8583ef7f0
galaxyp
parents:
diff changeset
51 args = parser.parse_args()
887e111c0919 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/proteogenomics/retrieve_ensembl_bed commit 3fd7be931712e7fa5b281bc8c48104c8583ef7f0
galaxyp
parents:
diff changeset
52 # print >> sys.stderr, "args: %s" % args
887e111c0919 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/proteogenomics/retrieve_ensembl_bed commit 3fd7be931712e7fa5b281bc8c48104c8583ef7f0
galaxyp
parents:
diff changeset
53 species = args.species
887e111c0919 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/proteogenomics/retrieve_ensembl_bed commit 3fd7be931712e7fa5b281bc8c48104c8583ef7f0
galaxyp
parents:
diff changeset
54 out_wtr = open(args.output, 'w') if args.output != '-' else sys.stdout
887e111c0919 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/proteogenomics/retrieve_ensembl_bed commit 3fd7be931712e7fa5b281bc8c48104c8583ef7f0
galaxyp
parents:
diff changeset
55 biotypes = ';'.join(['biotype=%s' % bt.strip()
887e111c0919 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/proteogenomics/retrieve_ensembl_bed commit 3fd7be931712e7fa5b281bc8c48104c8583ef7f0
galaxyp
parents:
diff changeset
56 for biotype in args.biotypes
887e111c0919 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/proteogenomics/retrieve_ensembl_bed commit 3fd7be931712e7fa5b281bc8c48104c8583ef7f0
galaxyp
parents:
diff changeset
57 for bt in biotype.split(',') if bt.strip()])
887e111c0919 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/proteogenomics/retrieve_ensembl_bed commit 3fd7be931712e7fa5b281bc8c48104c8583ef7f0
galaxyp
parents:
diff changeset
58
887e111c0919 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/proteogenomics/retrieve_ensembl_bed commit 3fd7be931712e7fa5b281bc8c48104c8583ef7f0
galaxyp
parents:
diff changeset
59 selected_regions = dict() # chrom:(start, end)
887e111c0919 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/proteogenomics/retrieve_ensembl_bed commit 3fd7be931712e7fa5b281bc8c48104c8583ef7f0
galaxyp
parents:
diff changeset
60 region_pat = '^([^:]+)(?::(\d*)(?:-(\d+)([+-])?)?)?'
887e111c0919 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/proteogenomics/retrieve_ensembl_bed commit 3fd7be931712e7fa5b281bc8c48104c8583ef7f0
galaxyp
parents:
diff changeset
61 if args.regions:
887e111c0919 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/proteogenomics/retrieve_ensembl_bed commit 3fd7be931712e7fa5b281bc8c48104c8583ef7f0
galaxyp
parents:
diff changeset
62 for entry in args.regions:
887e111c0919 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/proteogenomics/retrieve_ensembl_bed commit 3fd7be931712e7fa5b281bc8c48104c8583ef7f0
galaxyp
parents:
diff changeset
63 if not entry:
887e111c0919 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/proteogenomics/retrieve_ensembl_bed commit 3fd7be931712e7fa5b281bc8c48104c8583ef7f0
galaxyp
parents:
diff changeset
64 continue
887e111c0919 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/proteogenomics/retrieve_ensembl_bed commit 3fd7be931712e7fa5b281bc8c48104c8583ef7f0
galaxyp
parents:
diff changeset
65 regs = [x.strip() for x in entry.split(',') if x.strip()]
887e111c0919 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/proteogenomics/retrieve_ensembl_bed commit 3fd7be931712e7fa5b281bc8c48104c8583ef7f0
galaxyp
parents:
diff changeset
66 for reg in regs:
887e111c0919 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/proteogenomics/retrieve_ensembl_bed commit 3fd7be931712e7fa5b281bc8c48104c8583ef7f0
galaxyp
parents:
diff changeset
67 m = re.match(region_pat, reg)
887e111c0919 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/proteogenomics/retrieve_ensembl_bed commit 3fd7be931712e7fa5b281bc8c48104c8583ef7f0
galaxyp
parents:
diff changeset
68 if m:
887e111c0919 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/proteogenomics/retrieve_ensembl_bed commit 3fd7be931712e7fa5b281bc8c48104c8583ef7f0
galaxyp
parents:
diff changeset
69 (chrom, start, end, strand) = m.groups()
887e111c0919 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/proteogenomics/retrieve_ensembl_bed commit 3fd7be931712e7fa5b281bc8c48104c8583ef7f0
galaxyp
parents:
diff changeset
70 if chrom:
887e111c0919 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/proteogenomics/retrieve_ensembl_bed commit 3fd7be931712e7fa5b281bc8c48104c8583ef7f0
galaxyp
parents:
diff changeset
71 if chrom not in selected_regions:
887e111c0919 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/proteogenomics/retrieve_ensembl_bed commit 3fd7be931712e7fa5b281bc8c48104c8583ef7f0
galaxyp
parents:
diff changeset
72 selected_regions[chrom] = []
887e111c0919 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/proteogenomics/retrieve_ensembl_bed commit 3fd7be931712e7fa5b281bc8c48104c8583ef7f0
galaxyp
parents:
diff changeset
73 selected_regions[chrom].append([start, end, strand])
887e111c0919 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/proteogenomics/retrieve_ensembl_bed commit 3fd7be931712e7fa5b281bc8c48104c8583ef7f0
galaxyp
parents:
diff changeset
74 if args.debug:
887e111c0919 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/proteogenomics/retrieve_ensembl_bed commit 3fd7be931712e7fa5b281bc8c48104c8583ef7f0
galaxyp
parents:
diff changeset
75 print >> sys.stderr, "selected_regions: %s" % selected_regions
887e111c0919 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/proteogenomics/retrieve_ensembl_bed commit 3fd7be931712e7fa5b281bc8c48104c8583ef7f0
galaxyp
parents:
diff changeset
76
887e111c0919 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/proteogenomics/retrieve_ensembl_bed commit 3fd7be931712e7fa5b281bc8c48104c8583ef7f0
galaxyp
parents:
diff changeset
77 def retrieve_region(species, ref, start, stop, strand):
887e111c0919 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/proteogenomics/retrieve_ensembl_bed commit 3fd7be931712e7fa5b281bc8c48104c8583ef7f0
galaxyp
parents:
diff changeset
78 transcript_count = 0
887e111c0919 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/proteogenomics/retrieve_ensembl_bed commit 3fd7be931712e7fa5b281bc8c48104c8583ef7f0
galaxyp
parents:
diff changeset
79 regions = range(start, stop, max_region)
887e111c0919 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/proteogenomics/retrieve_ensembl_bed commit 3fd7be931712e7fa5b281bc8c48104c8583ef7f0
galaxyp
parents:
diff changeset
80 if not regions or regions[-1] < stop:
887e111c0919 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/proteogenomics/retrieve_ensembl_bed commit 3fd7be931712e7fa5b281bc8c48104c8583ef7f0
galaxyp
parents:
diff changeset
81 regions.append(stop)
887e111c0919 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/proteogenomics/retrieve_ensembl_bed commit 3fd7be931712e7fa5b281bc8c48104c8583ef7f0
galaxyp
parents:
diff changeset
82 for end in regions[1:]:
887e111c0919 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/proteogenomics/retrieve_ensembl_bed commit 3fd7be931712e7fa5b281bc8c48104c8583ef7f0
galaxyp
parents:
diff changeset
83 bedlines = get_transcripts_bed(species, ref, start, end,
887e111c0919 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/proteogenomics/retrieve_ensembl_bed commit 3fd7be931712e7fa5b281bc8c48104c8583ef7f0
galaxyp
parents:
diff changeset
84 strand=strand, params=biotypes)
887e111c0919 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/proteogenomics/retrieve_ensembl_bed commit 3fd7be931712e7fa5b281bc8c48104c8583ef7f0
galaxyp
parents:
diff changeset
85 if args.debug:
887e111c0919 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/proteogenomics/retrieve_ensembl_bed commit 3fd7be931712e7fa5b281bc8c48104c8583ef7f0
galaxyp
parents:
diff changeset
86 print >> sys.stderr,\
887e111c0919 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/proteogenomics/retrieve_ensembl_bed commit 3fd7be931712e7fa5b281bc8c48104c8583ef7f0
galaxyp
parents:
diff changeset
87 "%s\t%s\tstart: %d\tend: %d\tcDNA transcripts:%d"\
887e111c0919 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/proteogenomics/retrieve_ensembl_bed commit 3fd7be931712e7fa5b281bc8c48104c8583ef7f0
galaxyp
parents:
diff changeset
88 % (species, ref, start, end, len(bedlines))
887e111c0919 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/proteogenomics/retrieve_ensembl_bed commit 3fd7be931712e7fa5b281bc8c48104c8583ef7f0
galaxyp
parents:
diff changeset
89 # start, end, seq
887e111c0919 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/proteogenomics/retrieve_ensembl_bed commit 3fd7be931712e7fa5b281bc8c48104c8583ef7f0
galaxyp
parents:
diff changeset
90 for i, bedline in enumerate(bedlines):
887e111c0919 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/proteogenomics/retrieve_ensembl_bed commit 3fd7be931712e7fa5b281bc8c48104c8583ef7f0
galaxyp
parents:
diff changeset
91 if args.debug:
887e111c0919 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/proteogenomics/retrieve_ensembl_bed commit 3fd7be931712e7fa5b281bc8c48104c8583ef7f0
galaxyp
parents:
diff changeset
92 print >> sys.stderr, "%s\n" % (bedline)
887e111c0919 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/proteogenomics/retrieve_ensembl_bed commit 3fd7be931712e7fa5b281bc8c48104c8583ef7f0
galaxyp
parents:
diff changeset
93 if not args.ucsc_chrom_names:
887e111c0919 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/proteogenomics/retrieve_ensembl_bed commit 3fd7be931712e7fa5b281bc8c48104c8583ef7f0
galaxyp
parents:
diff changeset
94 bedline = re.sub('^[^\t]+', ref, bedline)
887e111c0919 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/proteogenomics/retrieve_ensembl_bed commit 3fd7be931712e7fa5b281bc8c48104c8583ef7f0
galaxyp
parents:
diff changeset
95 try:
887e111c0919 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/proteogenomics/retrieve_ensembl_bed commit 3fd7be931712e7fa5b281bc8c48104c8583ef7f0
galaxyp
parents:
diff changeset
96 if out_wtr:
887e111c0919 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/proteogenomics/retrieve_ensembl_bed commit 3fd7be931712e7fa5b281bc8c48104c8583ef7f0
galaxyp
parents:
diff changeset
97 out_wtr.write(bedline.replace(',\t', '\t')
887e111c0919 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/proteogenomics/retrieve_ensembl_bed commit 3fd7be931712e7fa5b281bc8c48104c8583ef7f0
galaxyp
parents:
diff changeset
98 if args.extended_bed
887e111c0919 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/proteogenomics/retrieve_ensembl_bed commit 3fd7be931712e7fa5b281bc8c48104c8583ef7f0
galaxyp
parents:
diff changeset
99 else str(bed_from_line(bedline)))
887e111c0919 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/proteogenomics/retrieve_ensembl_bed commit 3fd7be931712e7fa5b281bc8c48104c8583ef7f0
galaxyp
parents:
diff changeset
100 out_wtr.write("\n")
887e111c0919 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/proteogenomics/retrieve_ensembl_bed commit 3fd7be931712e7fa5b281bc8c48104c8583ef7f0
galaxyp
parents:
diff changeset
101 out_wtr.flush()
887e111c0919 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/proteogenomics/retrieve_ensembl_bed commit 3fd7be931712e7fa5b281bc8c48104c8583ef7f0
galaxyp
parents:
diff changeset
102 except Exception as e:
887e111c0919 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/proteogenomics/retrieve_ensembl_bed commit 3fd7be931712e7fa5b281bc8c48104c8583ef7f0
galaxyp
parents:
diff changeset
103 print >> sys.stderr,\
887e111c0919 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/proteogenomics/retrieve_ensembl_bed commit 3fd7be931712e7fa5b281bc8c48104c8583ef7f0
galaxyp
parents:
diff changeset
104 "BED error (%s) : %s\n" % (e, bedline)
887e111c0919 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/proteogenomics/retrieve_ensembl_bed commit 3fd7be931712e7fa5b281bc8c48104c8583ef7f0
galaxyp
parents:
diff changeset
105 start = end + 1
887e111c0919 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/proteogenomics/retrieve_ensembl_bed commit 3fd7be931712e7fa5b281bc8c48104c8583ef7f0
galaxyp
parents:
diff changeset
106 return transcript_count
887e111c0919 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/proteogenomics/retrieve_ensembl_bed commit 3fd7be931712e7fa5b281bc8c48104c8583ef7f0
galaxyp
parents:
diff changeset
107
887e111c0919 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/proteogenomics/retrieve_ensembl_bed commit 3fd7be931712e7fa5b281bc8c48104c8583ef7f0
galaxyp
parents:
diff changeset
108 coord_systems = get_toplevel(species)
887e111c0919 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/proteogenomics/retrieve_ensembl_bed commit 3fd7be931712e7fa5b281bc8c48104c8583ef7f0
galaxyp
parents:
diff changeset
109 if 'chromosome' in coord_systems:
887e111c0919 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/proteogenomics/retrieve_ensembl_bed commit 3fd7be931712e7fa5b281bc8c48104c8583ef7f0
galaxyp
parents:
diff changeset
110 ref_lengths = dict()
887e111c0919 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/proteogenomics/retrieve_ensembl_bed commit 3fd7be931712e7fa5b281bc8c48104c8583ef7f0
galaxyp
parents:
diff changeset
111 for ref in sorted(coord_systems['chromosome'].keys()):
887e111c0919 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/proteogenomics/retrieve_ensembl_bed commit 3fd7be931712e7fa5b281bc8c48104c8583ef7f0
galaxyp
parents:
diff changeset
112 length = coord_systems['chromosome'][ref]
887e111c0919 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/proteogenomics/retrieve_ensembl_bed commit 3fd7be931712e7fa5b281bc8c48104c8583ef7f0
galaxyp
parents:
diff changeset
113 ref_lengths[ref] = length
887e111c0919 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/proteogenomics/retrieve_ensembl_bed commit 3fd7be931712e7fa5b281bc8c48104c8583ef7f0
galaxyp
parents:
diff changeset
114 if args.toplevel:
887e111c0919 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/proteogenomics/retrieve_ensembl_bed commit 3fd7be931712e7fa5b281bc8c48104c8583ef7f0
galaxyp
parents:
diff changeset
115 print >> sys.stderr,\
887e111c0919 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/proteogenomics/retrieve_ensembl_bed commit 3fd7be931712e7fa5b281bc8c48104c8583ef7f0
galaxyp
parents:
diff changeset
116 "%s\t%s\tlength: %d" % (species, ref, length)
887e111c0919 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/proteogenomics/retrieve_ensembl_bed commit 3fd7be931712e7fa5b281bc8c48104c8583ef7f0
galaxyp
parents:
diff changeset
117 if selected_regions:
887e111c0919 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/proteogenomics/retrieve_ensembl_bed commit 3fd7be931712e7fa5b281bc8c48104c8583ef7f0
galaxyp
parents:
diff changeset
118 transcript_count = 0
887e111c0919 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/proteogenomics/retrieve_ensembl_bed commit 3fd7be931712e7fa5b281bc8c48104c8583ef7f0
galaxyp
parents:
diff changeset
119 for ref in sorted(selected_regions.keys()):
887e111c0919 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/proteogenomics/retrieve_ensembl_bed commit 3fd7be931712e7fa5b281bc8c48104c8583ef7f0
galaxyp
parents:
diff changeset
120 if ref in ref_lengths:
887e111c0919 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/proteogenomics/retrieve_ensembl_bed commit 3fd7be931712e7fa5b281bc8c48104c8583ef7f0
galaxyp
parents:
diff changeset
121 for reg in selected_regions[ref]:
887e111c0919 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/proteogenomics/retrieve_ensembl_bed commit 3fd7be931712e7fa5b281bc8c48104c8583ef7f0
galaxyp
parents:
diff changeset
122 (_start, _stop, _strand) = reg
887e111c0919 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/proteogenomics/retrieve_ensembl_bed commit 3fd7be931712e7fa5b281bc8c48104c8583ef7f0
galaxyp
parents:
diff changeset
123 start = int(_start) if _start else 0
887e111c0919 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/proteogenomics/retrieve_ensembl_bed commit 3fd7be931712e7fa5b281bc8c48104c8583ef7f0
galaxyp
parents:
diff changeset
124 stop = int(_stop) if _stop else ref_lengths[ref]
887e111c0919 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/proteogenomics/retrieve_ensembl_bed commit 3fd7be931712e7fa5b281bc8c48104c8583ef7f0
galaxyp
parents:
diff changeset
125 strand = '' if not _strand else ':1'\
887e111c0919 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/proteogenomics/retrieve_ensembl_bed commit 3fd7be931712e7fa5b281bc8c48104c8583ef7f0
galaxyp
parents:
diff changeset
126 if _strand == '+' else ':-1'
887e111c0919 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/proteogenomics/retrieve_ensembl_bed commit 3fd7be931712e7fa5b281bc8c48104c8583ef7f0
galaxyp
parents:
diff changeset
127 transcript_count += retrieve_region(species, ref,
887e111c0919 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/proteogenomics/retrieve_ensembl_bed commit 3fd7be931712e7fa5b281bc8c48104c8583ef7f0
galaxyp
parents:
diff changeset
128 start, stop,
887e111c0919 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/proteogenomics/retrieve_ensembl_bed commit 3fd7be931712e7fa5b281bc8c48104c8583ef7f0
galaxyp
parents:
diff changeset
129 strand)
887e111c0919 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/proteogenomics/retrieve_ensembl_bed commit 3fd7be931712e7fa5b281bc8c48104c8583ef7f0
galaxyp
parents:
diff changeset
130 if args.debug or args.verbose:
887e111c0919 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/proteogenomics/retrieve_ensembl_bed commit 3fd7be931712e7fa5b281bc8c48104c8583ef7f0
galaxyp
parents:
diff changeset
131 length = stop - start
887e111c0919 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/proteogenomics/retrieve_ensembl_bed commit 3fd7be931712e7fa5b281bc8c48104c8583ef7f0
galaxyp
parents:
diff changeset
132 print >> sys.stderr,\
887e111c0919 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/proteogenomics/retrieve_ensembl_bed commit 3fd7be931712e7fa5b281bc8c48104c8583ef7f0
galaxyp
parents:
diff changeset
133 "%s\t%s:%d-%d%s\tlength: %d\ttrancripts:%d"\
887e111c0919 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/proteogenomics/retrieve_ensembl_bed commit 3fd7be931712e7fa5b281bc8c48104c8583ef7f0
galaxyp
parents:
diff changeset
134 % (species, ref, start, stop, strand,
887e111c0919 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/proteogenomics/retrieve_ensembl_bed commit 3fd7be931712e7fa5b281bc8c48104c8583ef7f0
galaxyp
parents:
diff changeset
135 length, transcript_count)
887e111c0919 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/proteogenomics/retrieve_ensembl_bed commit 3fd7be931712e7fa5b281bc8c48104c8583ef7f0
galaxyp
parents:
diff changeset
136 else:
887e111c0919 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/proteogenomics/retrieve_ensembl_bed commit 3fd7be931712e7fa5b281bc8c48104c8583ef7f0
galaxyp
parents:
diff changeset
137 strand = ''
887e111c0919 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/proteogenomics/retrieve_ensembl_bed commit 3fd7be931712e7fa5b281bc8c48104c8583ef7f0
galaxyp
parents:
diff changeset
138 start = 0
887e111c0919 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/proteogenomics/retrieve_ensembl_bed commit 3fd7be931712e7fa5b281bc8c48104c8583ef7f0
galaxyp
parents:
diff changeset
139 for ref in sorted(ref_lengths.keys()):
887e111c0919 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/proteogenomics/retrieve_ensembl_bed commit 3fd7be931712e7fa5b281bc8c48104c8583ef7f0
galaxyp
parents:
diff changeset
140 length = ref_lengths[ref]
887e111c0919 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/proteogenomics/retrieve_ensembl_bed commit 3fd7be931712e7fa5b281bc8c48104c8583ef7f0
galaxyp
parents:
diff changeset
141 transcript_count = 0
887e111c0919 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/proteogenomics/retrieve_ensembl_bed commit 3fd7be931712e7fa5b281bc8c48104c8583ef7f0
galaxyp
parents:
diff changeset
142 if args.debug:
887e111c0919 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/proteogenomics/retrieve_ensembl_bed commit 3fd7be931712e7fa5b281bc8c48104c8583ef7f0
galaxyp
parents:
diff changeset
143 print >> sys.stderr,\
887e111c0919 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/proteogenomics/retrieve_ensembl_bed commit 3fd7be931712e7fa5b281bc8c48104c8583ef7f0
galaxyp
parents:
diff changeset
144 "Retrieving transcripts: %s\t%s\tlength: %d"\
887e111c0919 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/proteogenomics/retrieve_ensembl_bed commit 3fd7be931712e7fa5b281bc8c48104c8583ef7f0
galaxyp
parents:
diff changeset
145 % (species, ref, length)
887e111c0919 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/proteogenomics/retrieve_ensembl_bed commit 3fd7be931712e7fa5b281bc8c48104c8583ef7f0
galaxyp
parents:
diff changeset
146 transcript_count += retrieve_region(species, ref, start,
887e111c0919 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/proteogenomics/retrieve_ensembl_bed commit 3fd7be931712e7fa5b281bc8c48104c8583ef7f0
galaxyp
parents:
diff changeset
147 length, strand)
887e111c0919 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/proteogenomics/retrieve_ensembl_bed commit 3fd7be931712e7fa5b281bc8c48104c8583ef7f0
galaxyp
parents:
diff changeset
148 if args.debug or args.verbose:
887e111c0919 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/proteogenomics/retrieve_ensembl_bed commit 3fd7be931712e7fa5b281bc8c48104c8583ef7f0
galaxyp
parents:
diff changeset
149 print >> sys.stderr,\
887e111c0919 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/proteogenomics/retrieve_ensembl_bed commit 3fd7be931712e7fa5b281bc8c48104c8583ef7f0
galaxyp
parents:
diff changeset
150 "%s\t%s\tlength: %d\ttrancripts:%d"\
887e111c0919 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/proteogenomics/retrieve_ensembl_bed commit 3fd7be931712e7fa5b281bc8c48104c8583ef7f0
galaxyp
parents:
diff changeset
151 % (species, ref, length, transcript_count)
887e111c0919 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/proteogenomics/retrieve_ensembl_bed commit 3fd7be931712e7fa5b281bc8c48104c8583ef7f0
galaxyp
parents:
diff changeset
152
887e111c0919 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/proteogenomics/retrieve_ensembl_bed commit 3fd7be931712e7fa5b281bc8c48104c8583ef7f0
galaxyp
parents:
diff changeset
153
887e111c0919 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/proteogenomics/retrieve_ensembl_bed commit 3fd7be931712e7fa5b281bc8c48104c8583ef7f0
galaxyp
parents:
diff changeset
154 if __name__ == "__main__":
887e111c0919 planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/proteogenomics/retrieve_ensembl_bed commit 3fd7be931712e7fa5b281bc8c48104c8583ef7f0
galaxyp
parents:
diff changeset
155 __main__()