Mercurial > repos > lldelisle > fromgtftobed12

annotate fromgtfTobed12.py @ 1:6fd4b3b90220 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/lldelisle/tools-lldelisle/tree/master/tools/fromgtfTobed12 commit 15b8c2cc83708044413a152322bcbfca8a74d29a
author lldelisle
date Fri, 03 Nov 2023 14:13:51 +0000
parents 418e4d0fe0bd
children
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
rev   line source
0
418e4d0fe0bd planemo upload for repository https://github.com/lldelisle/tools-lldelisle/tree/master/tools/fromgtfTobed12 commit 1aaffda5b95e0389e315179345642c0d005867c1
lldelisle
parents:
diff changeset
1 import argparse
418e4d0fe0bd planemo upload for repository https://github.com/lldelisle/tools-lldelisle/tree/master/tools/fromgtfTobed12 commit 1aaffda5b95e0389e315179345642c0d005867c1
lldelisle
parents:
diff changeset
2 import sys
418e4d0fe0bd planemo upload for repository https://github.com/lldelisle/tools-lldelisle/tree/master/tools/fromgtfTobed12 commit 1aaffda5b95e0389e315179345642c0d005867c1
lldelisle
parents:
diff changeset
3 import warnings
418e4d0fe0bd planemo upload for repository https://github.com/lldelisle/tools-lldelisle/tree/master/tools/fromgtfTobed12 commit 1aaffda5b95e0389e315179345642c0d005867c1
lldelisle
parents:
diff changeset
4
418e4d0fe0bd planemo upload for repository https://github.com/lldelisle/tools-lldelisle/tree/master/tools/fromgtfTobed12 commit 1aaffda5b95e0389e315179345642c0d005867c1
lldelisle
parents:
diff changeset
5 import gffutils
418e4d0fe0bd planemo upload for repository https://github.com/lldelisle/tools-lldelisle/tree/master/tools/fromgtfTobed12 commit 1aaffda5b95e0389e315179345642c0d005867c1
lldelisle
parents:
diff changeset
6
418e4d0fe0bd planemo upload for repository https://github.com/lldelisle/tools-lldelisle/tree/master/tools/fromgtfTobed12 commit 1aaffda5b95e0389e315179345642c0d005867c1
lldelisle
parents:
diff changeset
7 warnings.filterwarnings("ignore", message="It appears you have a gene feature"
418e4d0fe0bd planemo upload for repository https://github.com/lldelisle/tools-lldelisle/tree/master/tools/fromgtfTobed12 commit 1aaffda5b95e0389e315179345642c0d005867c1
lldelisle
parents:
diff changeset
8 " in your GTF file. You may want to use the "
418e4d0fe0bd planemo upload for repository https://github.com/lldelisle/tools-lldelisle/tree/master/tools/fromgtfTobed12 commit 1aaffda5b95e0389e315179345642c0d005867c1
lldelisle
parents:
diff changeset
9 "`disable_infer_genes` option to speed up database "
418e4d0fe0bd planemo upload for repository https://github.com/lldelisle/tools-lldelisle/tree/master/tools/fromgtfTobed12 commit 1aaffda5b95e0389e315179345642c0d005867c1
lldelisle
parents:
diff changeset
10 "creation")
418e4d0fe0bd planemo upload for repository https://github.com/lldelisle/tools-lldelisle/tree/master/tools/fromgtfTobed12 commit 1aaffda5b95e0389e315179345642c0d005867c1
lldelisle
parents:
diff changeset
11 warnings.filterwarnings("ignore", message="It appears you have a transcript "
418e4d0fe0bd planemo upload for repository https://github.com/lldelisle/tools-lldelisle/tree/master/tools/fromgtfTobed12 commit 1aaffda5b95e0389e315179345642c0d005867c1
lldelisle
parents:
diff changeset
12 "feature in your GTF file. You may want to use the "
418e4d0fe0bd planemo upload for repository https://github.com/lldelisle/tools-lldelisle/tree/master/tools/fromgtfTobed12 commit 1aaffda5b95e0389e315179345642c0d005867c1
lldelisle
parents:
diff changeset
13 "`disable_infer_transcripts` option to speed up "
418e4d0fe0bd planemo upload for repository https://github.com/lldelisle/tools-lldelisle/tree/master/tools/fromgtfTobed12 commit 1aaffda5b95e0389e315179345642c0d005867c1
lldelisle
parents:
diff changeset
14 "database creation")
418e4d0fe0bd planemo upload for repository https://github.com/lldelisle/tools-lldelisle/tree/master/tools/fromgtfTobed12 commit 1aaffda5b95e0389e315179345642c0d005867c1
lldelisle
parents:
diff changeset
15 # In gffutils v0.10 they changed the error message:
418e4d0fe0bd planemo upload for repository https://github.com/lldelisle/tools-lldelisle/tree/master/tools/fromgtfTobed12 commit 1aaffda5b95e0389e315179345642c0d005867c1
lldelisle
parents:
diff changeset
16 warnings.filterwarnings("ignore", message="It appears you have a gene feature"
418e4d0fe0bd planemo upload for repository https://github.com/lldelisle/tools-lldelisle/tree/master/tools/fromgtfTobed12 commit 1aaffda5b95e0389e315179345642c0d005867c1
lldelisle
parents:
diff changeset
17 " in your GTF file. You may want to use the "
418e4d0fe0bd planemo upload for repository https://github.com/lldelisle/tools-lldelisle/tree/master/tools/fromgtfTobed12 commit 1aaffda5b95e0389e315179345642c0d005867c1
lldelisle
parents:
diff changeset
18 "`disable_infer_genes=True` option to speed up "
418e4d0fe0bd planemo upload for repository https://github.com/lldelisle/tools-lldelisle/tree/master/tools/fromgtfTobed12 commit 1aaffda5b95e0389e315179345642c0d005867c1
lldelisle
parents:
diff changeset
19 "database creation")
418e4d0fe0bd planemo upload for repository https://github.com/lldelisle/tools-lldelisle/tree/master/tools/fromgtfTobed12 commit 1aaffda5b95e0389e315179345642c0d005867c1
lldelisle
parents:
diff changeset
20 warnings.filterwarnings("ignore", message="It appears you have a transcript "
418e4d0fe0bd planemo upload for repository https://github.com/lldelisle/tools-lldelisle/tree/master/tools/fromgtfTobed12 commit 1aaffda5b95e0389e315179345642c0d005867c1
lldelisle
parents:
diff changeset
21 "feature in your GTF file. You may want to use the "
418e4d0fe0bd planemo upload for repository https://github.com/lldelisle/tools-lldelisle/tree/master/tools/fromgtfTobed12 commit 1aaffda5b95e0389e315179345642c0d005867c1
lldelisle
parents:
diff changeset
22 "`disable_infer_transcripts=True` option to speed up "
418e4d0fe0bd planemo upload for repository https://github.com/lldelisle/tools-lldelisle/tree/master/tools/fromgtfTobed12 commit 1aaffda5b95e0389e315179345642c0d005867c1
lldelisle
parents:
diff changeset
23 "database creation")
418e4d0fe0bd planemo upload for repository https://github.com/lldelisle/tools-lldelisle/tree/master/tools/fromgtfTobed12 commit 1aaffda5b95e0389e315179345642c0d005867c1
lldelisle
parents:
diff changeset
24
418e4d0fe0bd planemo upload for repository https://github.com/lldelisle/tools-lldelisle/tree/master/tools/fromgtfTobed12 commit 1aaffda5b95e0389e315179345642c0d005867c1
lldelisle
parents:
diff changeset
25
1
6fd4b3b90220 planemo upload for repository https://github.com/lldelisle/tools-lldelisle/tree/master/tools/fromgtfTobed12 commit 15b8c2cc83708044413a152322bcbfca8a74d29a
lldelisle
parents: 0
diff changeset
26 def convert_gtf_to_bed(fn, fo, preferedName, mergeTranscripts,
0
418e4d0fe0bd planemo upload for repository https://github.com/lldelisle/tools-lldelisle/tree/master/tools/fromgtfTobed12 commit 1aaffda5b95e0389e315179345642c0d005867c1
lldelisle
parents:
diff changeset
27 mergeTranscriptsAndOverlappingExons, ucsc):
418e4d0fe0bd planemo upload for repository https://github.com/lldelisle/tools-lldelisle/tree/master/tools/fromgtfTobed12 commit 1aaffda5b95e0389e315179345642c0d005867c1
lldelisle
parents:
diff changeset
28 db = gffutils.create_db(fn, ':memory:')
418e4d0fe0bd planemo upload for repository https://github.com/lldelisle/tools-lldelisle/tree/master/tools/fromgtfTobed12 commit 1aaffda5b95e0389e315179345642c0d005867c1
lldelisle
parents:
diff changeset
29 # For each transcript:
1
6fd4b3b90220 planemo upload for repository https://github.com/lldelisle/tools-lldelisle/tree/master/tools/fromgtfTobed12 commit 15b8c2cc83708044413a152322bcbfca8a74d29a
lldelisle
parents: 0
diff changeset
30 if preferedName is not None:
6fd4b3b90220 planemo upload for repository https://github.com/lldelisle/tools-lldelisle/tree/master/tools/fromgtfTobed12 commit 15b8c2cc83708044413a152322bcbfca8a74d29a
lldelisle
parents: 0
diff changeset
31 prefered_name = preferedName
6fd4b3b90220 planemo upload for repository https://github.com/lldelisle/tools-lldelisle/tree/master/tools/fromgtfTobed12 commit 15b8c2cc83708044413a152322bcbfca8a74d29a
lldelisle
parents: 0
diff changeset
32 elif mergeTranscripts or mergeTranscriptsAndOverlappingExons:
0
418e4d0fe0bd planemo upload for repository https://github.com/lldelisle/tools-lldelisle/tree/master/tools/fromgtfTobed12 commit 1aaffda5b95e0389e315179345642c0d005867c1
lldelisle
parents:
diff changeset
33 prefered_name = "gene_name"
1
6fd4b3b90220 planemo upload for repository https://github.com/lldelisle/tools-lldelisle/tree/master/tools/fromgtfTobed12 commit 15b8c2cc83708044413a152322bcbfca8a74d29a
lldelisle
parents: 0
diff changeset
34 else:
6fd4b3b90220 planemo upload for repository https://github.com/lldelisle/tools-lldelisle/tree/master/tools/fromgtfTobed12 commit 15b8c2cc83708044413a152322bcbfca8a74d29a
lldelisle
parents: 0
diff changeset
35 prefered_name = "transcript_name"
0
418e4d0fe0bd planemo upload for repository https://github.com/lldelisle/tools-lldelisle/tree/master/tools/fromgtfTobed12 commit 1aaffda5b95e0389e315179345642c0d005867c1
lldelisle
parents:
diff changeset
36 if mergeTranscripts or mergeTranscriptsAndOverlappingExons:
418e4d0fe0bd planemo upload for repository https://github.com/lldelisle/tools-lldelisle/tree/master/tools/fromgtfTobed12 commit 1aaffda5b95e0389e315179345642c0d005867c1
lldelisle
parents:
diff changeset
37 all_items = db.features_of_type("gene", order_by='start')
418e4d0fe0bd planemo upload for repository https://github.com/lldelisle/tools-lldelisle/tree/master/tools/fromgtfTobed12 commit 1aaffda5b95e0389e315179345642c0d005867c1
lldelisle
parents:
diff changeset
38 else:
418e4d0fe0bd planemo upload for repository https://github.com/lldelisle/tools-lldelisle/tree/master/tools/fromgtfTobed12 commit 1aaffda5b95e0389e315179345642c0d005867c1
lldelisle
parents:
diff changeset
39 all_items = db.features_of_type("transcript", order_by='start')
418e4d0fe0bd planemo upload for repository https://github.com/lldelisle/tools-lldelisle/tree/master/tools/fromgtfTobed12 commit 1aaffda5b95e0389e315179345642c0d005867c1
lldelisle
parents:
diff changeset
40 for tr in all_items:
418e4d0fe0bd planemo upload for repository https://github.com/lldelisle/tools-lldelisle/tree/master/tools/fromgtfTobed12 commit 1aaffda5b95e0389e315179345642c0d005867c1
lldelisle
parents:
diff changeset
41 # The name would be the name of the transcript/gene if exists
418e4d0fe0bd planemo upload for repository https://github.com/lldelisle/tools-lldelisle/tree/master/tools/fromgtfTobed12 commit 1aaffda5b95e0389e315179345642c0d005867c1
lldelisle
parents:
diff changeset
42 try:
418e4d0fe0bd planemo upload for repository https://github.com/lldelisle/tools-lldelisle/tree/master/tools/fromgtfTobed12 commit 1aaffda5b95e0389e315179345642c0d005867c1
lldelisle
parents:
diff changeset
43 # First try to have it directly on the feature
418e4d0fe0bd planemo upload for repository https://github.com/lldelisle/tools-lldelisle/tree/master/tools/fromgtfTobed12 commit 1aaffda5b95e0389e315179345642c0d005867c1
lldelisle
parents:
diff changeset
44 trName = tr.attributes[prefered_name][0]
418e4d0fe0bd planemo upload for repository https://github.com/lldelisle/tools-lldelisle/tree/master/tools/fromgtfTobed12 commit 1aaffda5b95e0389e315179345642c0d005867c1
lldelisle
parents:
diff changeset
45 except KeyError:
418e4d0fe0bd planemo upload for repository https://github.com/lldelisle/tools-lldelisle/tree/master/tools/fromgtfTobed12 commit 1aaffda5b95e0389e315179345642c0d005867c1
lldelisle
parents:
diff changeset
46 # Else try to guess the name of the transcript/gene from exons:
418e4d0fe0bd planemo upload for repository https://github.com/lldelisle/tools-lldelisle/tree/master/tools/fromgtfTobed12 commit 1aaffda5b95e0389e315179345642c0d005867c1
lldelisle
parents:
diff changeset
47 try:
418e4d0fe0bd planemo upload for repository https://github.com/lldelisle/tools-lldelisle/tree/master/tools/fromgtfTobed12 commit 1aaffda5b95e0389e315179345642c0d005867c1
lldelisle
parents:
diff changeset
48 trName = set([e.attributes[prefered_name][0]
418e4d0fe0bd planemo upload for repository https://github.com/lldelisle/tools-lldelisle/tree/master/tools/fromgtfTobed12 commit 1aaffda5b95e0389e315179345642c0d005867c1
lldelisle
parents:
diff changeset
49 for e in
418e4d0fe0bd planemo upload for repository https://github.com/lldelisle/tools-lldelisle/tree/master/tools/fromgtfTobed12 commit 1aaffda5b95e0389e315179345642c0d005867c1
lldelisle
parents:
diff changeset
50 db.children(tr,
418e4d0fe0bd planemo upload for repository https://github.com/lldelisle/tools-lldelisle/tree/master/tools/fromgtfTobed12 commit 1aaffda5b95e0389e315179345642c0d005867c1
lldelisle
parents:
diff changeset
51 featuretype='exon',
418e4d0fe0bd planemo upload for repository https://github.com/lldelisle/tools-lldelisle/tree/master/tools/fromgtfTobed12 commit 1aaffda5b95e0389e315179345642c0d005867c1
lldelisle
parents:
diff changeset
52 order_by='start')]).pop()
418e4d0fe0bd planemo upload for repository https://github.com/lldelisle/tools-lldelisle/tree/master/tools/fromgtfTobed12 commit 1aaffda5b95e0389e315179345642c0d005867c1
lldelisle
parents:
diff changeset
53 except KeyError:
418e4d0fe0bd planemo upload for repository https://github.com/lldelisle/tools-lldelisle/tree/master/tools/fromgtfTobed12 commit 1aaffda5b95e0389e315179345642c0d005867c1
lldelisle
parents:
diff changeset
54 # Else take the transcript id
418e4d0fe0bd planemo upload for repository https://github.com/lldelisle/tools-lldelisle/tree/master/tools/fromgtfTobed12 commit 1aaffda5b95e0389e315179345642c0d005867c1
lldelisle
parents:
diff changeset
55 trName = tr.id
418e4d0fe0bd planemo upload for repository https://github.com/lldelisle/tools-lldelisle/tree/master/tools/fromgtfTobed12 commit 1aaffda5b95e0389e315179345642c0d005867c1
lldelisle
parents:
diff changeset
56 # If the cds is defined in the gtf,
418e4d0fe0bd planemo upload for repository https://github.com/lldelisle/tools-lldelisle/tree/master/tools/fromgtfTobed12 commit 1aaffda5b95e0389e315179345642c0d005867c1
lldelisle
parents:
diff changeset
57 # use it to define the thick start and end
418e4d0fe0bd planemo upload for repository https://github.com/lldelisle/tools-lldelisle/tree/master/tools/fromgtfTobed12 commit 1aaffda5b95e0389e315179345642c0d005867c1
lldelisle
parents:
diff changeset
58 # The gtf is 1-based closed intervalls and
418e4d0fe0bd planemo upload for repository https://github.com/lldelisle/tools-lldelisle/tree/master/tools/fromgtfTobed12 commit 1aaffda5b95e0389e315179345642c0d005867c1
lldelisle
parents:
diff changeset
59 # bed are 0-based half-open so:
418e4d0fe0bd planemo upload for repository https://github.com/lldelisle/tools-lldelisle/tree/master/tools/fromgtfTobed12 commit 1aaffda5b95e0389e315179345642c0d005867c1
lldelisle
parents:
diff changeset
60 # I need to remove one from each start
418e4d0fe0bd planemo upload for repository https://github.com/lldelisle/tools-lldelisle/tree/master/tools/fromgtfTobed12 commit 1aaffda5b95e0389e315179345642c0d005867c1
lldelisle
parents:
diff changeset
61 try:
418e4d0fe0bd planemo upload for repository https://github.com/lldelisle/tools-lldelisle/tree/master/tools/fromgtfTobed12 commit 1aaffda5b95e0389e315179345642c0d005867c1
lldelisle
parents:
diff changeset
62 # In case of multiple CDS (when there is one entry per gene)
418e4d0fe0bd planemo upload for repository https://github.com/lldelisle/tools-lldelisle/tree/master/tools/fromgtfTobed12 commit 1aaffda5b95e0389e315179345642c0d005867c1
lldelisle
parents:
diff changeset
63 # I use the first one to get the start
418e4d0fe0bd planemo upload for repository https://github.com/lldelisle/tools-lldelisle/tree/master/tools/fromgtfTobed12 commit 1aaffda5b95e0389e315179345642c0d005867c1
lldelisle
parents:
diff changeset
64 # and the last one to get the end (order_by=-start)
418e4d0fe0bd planemo upload for repository https://github.com/lldelisle/tools-lldelisle/tree/master/tools/fromgtfTobed12 commit 1aaffda5b95e0389e315179345642c0d005867c1
lldelisle
parents:
diff changeset
65 cds_start = next(db.children(tr,
418e4d0fe0bd planemo upload for repository https://github.com/lldelisle/tools-lldelisle/tree/master/tools/fromgtfTobed12 commit 1aaffda5b95e0389e315179345642c0d005867c1
lldelisle
parents:
diff changeset
66 featuretype='CDS',
418e4d0fe0bd planemo upload for repository https://github.com/lldelisle/tools-lldelisle/tree/master/tools/fromgtfTobed12 commit 1aaffda5b95e0389e315179345642c0d005867c1
lldelisle
parents:
diff changeset
67 order_by='start')).start - 1
418e4d0fe0bd planemo upload for repository https://github.com/lldelisle/tools-lldelisle/tree/master/tools/fromgtfTobed12 commit 1aaffda5b95e0389e315179345642c0d005867c1
lldelisle
parents:
diff changeset
68 cds_end = next(db.children(tr,
418e4d0fe0bd planemo upload for repository https://github.com/lldelisle/tools-lldelisle/tree/master/tools/fromgtfTobed12 commit 1aaffda5b95e0389e315179345642c0d005867c1
lldelisle
parents:
diff changeset
69 featuretype='CDS',
418e4d0fe0bd planemo upload for repository https://github.com/lldelisle/tools-lldelisle/tree/master/tools/fromgtfTobed12 commit 1aaffda5b95e0389e315179345642c0d005867c1
lldelisle
parents:
diff changeset
70 order_by='-start')).end
418e4d0fe0bd planemo upload for repository https://github.com/lldelisle/tools-lldelisle/tree/master/tools/fromgtfTobed12 commit 1aaffda5b95e0389e315179345642c0d005867c1
lldelisle
parents:
diff changeset
71 except StopIteration:
418e4d0fe0bd planemo upload for repository https://github.com/lldelisle/tools-lldelisle/tree/master/tools/fromgtfTobed12 commit 1aaffda5b95e0389e315179345642c0d005867c1
lldelisle
parents:
diff changeset
72 # If the CDS is not defined, then it is set to the start
418e4d0fe0bd planemo upload for repository https://github.com/lldelisle/tools-lldelisle/tree/master/tools/fromgtfTobed12 commit 1aaffda5b95e0389e315179345642c0d005867c1
lldelisle
parents:
diff changeset
73 # as proposed here:
418e4d0fe0bd planemo upload for repository https://github.com/lldelisle/tools-lldelisle/tree/master/tools/fromgtfTobed12 commit 1aaffda5b95e0389e315179345642c0d005867c1
lldelisle
parents:
diff changeset
74 # https://genome.ucsc.edu/FAQ/FAQformat.html#format1
418e4d0fe0bd planemo upload for repository https://github.com/lldelisle/tools-lldelisle/tree/master/tools/fromgtfTobed12 commit 1aaffda5b95e0389e315179345642c0d005867c1
lldelisle
parents:
diff changeset
75 cds_start = tr.start - 1
418e4d0fe0bd planemo upload for repository https://github.com/lldelisle/tools-lldelisle/tree/master/tools/fromgtfTobed12 commit 1aaffda5b95e0389e315179345642c0d005867c1
lldelisle
parents:
diff changeset
76 cds_end = tr.start - 1
418e4d0fe0bd planemo upload for repository https://github.com/lldelisle/tools-lldelisle/tree/master/tools/fromgtfTobed12 commit 1aaffda5b95e0389e315179345642c0d005867c1
lldelisle
parents:
diff changeset
77 # Get all exons starts and lengths
418e4d0fe0bd planemo upload for repository https://github.com/lldelisle/tools-lldelisle/tree/master/tools/fromgtfTobed12 commit 1aaffda5b95e0389e315179345642c0d005867c1
lldelisle
parents:
diff changeset
78 if mergeTranscriptsAndOverlappingExons:
418e4d0fe0bd planemo upload for repository https://github.com/lldelisle/tools-lldelisle/tree/master/tools/fromgtfTobed12 commit 1aaffda5b95e0389e315179345642c0d005867c1
lldelisle
parents:
diff changeset
79 # We merge overlapping exons:
418e4d0fe0bd planemo upload for repository https://github.com/lldelisle/tools-lldelisle/tree/master/tools/fromgtfTobed12 commit 1aaffda5b95e0389e315179345642c0d005867c1
lldelisle
parents:
diff changeset
80 exons_starts = []
418e4d0fe0bd planemo upload for repository https://github.com/lldelisle/tools-lldelisle/tree/master/tools/fromgtfTobed12 commit 1aaffda5b95e0389e315179345642c0d005867c1
lldelisle
parents:
diff changeset
81 exons_length = []
418e4d0fe0bd planemo upload for repository https://github.com/lldelisle/tools-lldelisle/tree/master/tools/fromgtfTobed12 commit 1aaffda5b95e0389e315179345642c0d005867c1
lldelisle
parents:
diff changeset
82 current_start = -1
418e4d0fe0bd planemo upload for repository https://github.com/lldelisle/tools-lldelisle/tree/master/tools/fromgtfTobed12 commit 1aaffda5b95e0389e315179345642c0d005867c1
lldelisle
parents:
diff changeset
83 current_end = None
418e4d0fe0bd planemo upload for repository https://github.com/lldelisle/tools-lldelisle/tree/master/tools/fromgtfTobed12 commit 1aaffda5b95e0389e315179345642c0d005867c1
lldelisle
parents:
diff changeset
84 for e in db.children(tr, featuretype='exon', order_by='start'):
418e4d0fe0bd planemo upload for repository https://github.com/lldelisle/tools-lldelisle/tree/master/tools/fromgtfTobed12 commit 1aaffda5b95e0389e315179345642c0d005867c1
lldelisle
parents:
diff changeset
85 if current_start == -1:
418e4d0fe0bd planemo upload for repository https://github.com/lldelisle/tools-lldelisle/tree/master/tools/fromgtfTobed12 commit 1aaffda5b95e0389e315179345642c0d005867c1
lldelisle
parents:
diff changeset
86 current_start = e.start - 1
418e4d0fe0bd planemo upload for repository https://github.com/lldelisle/tools-lldelisle/tree/master/tools/fromgtfTobed12 commit 1aaffda5b95e0389e315179345642c0d005867c1
lldelisle
parents:
diff changeset
87 current_end = e.end
418e4d0fe0bd planemo upload for repository https://github.com/lldelisle/tools-lldelisle/tree/master/tools/fromgtfTobed12 commit 1aaffda5b95e0389e315179345642c0d005867c1
lldelisle
parents:
diff changeset
88 else:
418e4d0fe0bd planemo upload for repository https://github.com/lldelisle/tools-lldelisle/tree/master/tools/fromgtfTobed12 commit 1aaffda5b95e0389e315179345642c0d005867c1
lldelisle
parents:
diff changeset
89 if e.start > current_end:
418e4d0fe0bd planemo upload for repository https://github.com/lldelisle/tools-lldelisle/tree/master/tools/fromgtfTobed12 commit 1aaffda5b95e0389e315179345642c0d005867c1
lldelisle
parents:
diff changeset
90 # This is a non-overlapping exon
418e4d0fe0bd planemo upload for repository https://github.com/lldelisle/tools-lldelisle/tree/master/tools/fromgtfTobed12 commit 1aaffda5b95e0389e315179345642c0d005867c1
lldelisle
parents:
diff changeset
91 # We store the previous exon:
418e4d0fe0bd planemo upload for repository https://github.com/lldelisle/tools-lldelisle/tree/master/tools/fromgtfTobed12 commit 1aaffda5b95e0389e315179345642c0d005867c1
lldelisle
parents:
diff changeset
92 exons_starts.append(current_start)
418e4d0fe0bd planemo upload for repository https://github.com/lldelisle/tools-lldelisle/tree/master/tools/fromgtfTobed12 commit 1aaffda5b95e0389e315179345642c0d005867c1
lldelisle
parents:
diff changeset
93 exons_length.append(current_end - current_start)
418e4d0fe0bd planemo upload for repository https://github.com/lldelisle/tools-lldelisle/tree/master/tools/fromgtfTobed12 commit 1aaffda5b95e0389e315179345642c0d005867c1
lldelisle
parents:
diff changeset
94 # We set the current:
418e4d0fe0bd planemo upload for repository https://github.com/lldelisle/tools-lldelisle/tree/master/tools/fromgtfTobed12 commit 1aaffda5b95e0389e315179345642c0d005867c1
lldelisle
parents:
diff changeset
95 current_start = e.start - 1
418e4d0fe0bd planemo upload for repository https://github.com/lldelisle/tools-lldelisle/tree/master/tools/fromgtfTobed12 commit 1aaffda5b95e0389e315179345642c0d005867c1
lldelisle
parents:
diff changeset
96 current_end = e.end
418e4d0fe0bd planemo upload for repository https://github.com/lldelisle/tools-lldelisle/tree/master/tools/fromgtfTobed12 commit 1aaffda5b95e0389e315179345642c0d005867c1
lldelisle
parents:
diff changeset
97 else:
418e4d0fe0bd planemo upload for repository https://github.com/lldelisle/tools-lldelisle/tree/master/tools/fromgtfTobed12 commit 1aaffda5b95e0389e315179345642c0d005867c1
lldelisle
parents:
diff changeset
98 # This is an overlapping exon
418e4d0fe0bd planemo upload for repository https://github.com/lldelisle/tools-lldelisle/tree/master/tools/fromgtfTobed12 commit 1aaffda5b95e0389e315179345642c0d005867c1
lldelisle
parents:
diff changeset
99 # We update current_end if necessary
418e4d0fe0bd planemo upload for repository https://github.com/lldelisle/tools-lldelisle/tree/master/tools/fromgtfTobed12 commit 1aaffda5b95e0389e315179345642c0d005867c1
lldelisle
parents:
diff changeset
100 current_end = max(current_end, e.end)
418e4d0fe0bd planemo upload for repository https://github.com/lldelisle/tools-lldelisle/tree/master/tools/fromgtfTobed12 commit 1aaffda5b95e0389e315179345642c0d005867c1
lldelisle
parents:
diff changeset
101 if current_start != -1:
418e4d0fe0bd planemo upload for repository https://github.com/lldelisle/tools-lldelisle/tree/master/tools/fromgtfTobed12 commit 1aaffda5b95e0389e315179345642c0d005867c1
lldelisle
parents:
diff changeset
102 # There is a last exon to store:
418e4d0fe0bd planemo upload for repository https://github.com/lldelisle/tools-lldelisle/tree/master/tools/fromgtfTobed12 commit 1aaffda5b95e0389e315179345642c0d005867c1
lldelisle
parents:
diff changeset
103 exons_starts.append(current_start)
418e4d0fe0bd planemo upload for repository https://github.com/lldelisle/tools-lldelisle/tree/master/tools/fromgtfTobed12 commit 1aaffda5b95e0389e315179345642c0d005867c1
lldelisle
parents:
diff changeset
104 exons_length.append(current_end - current_start)
418e4d0fe0bd planemo upload for repository https://github.com/lldelisle/tools-lldelisle/tree/master/tools/fromgtfTobed12 commit 1aaffda5b95e0389e315179345642c0d005867c1
lldelisle
parents:
diff changeset
105 else:
418e4d0fe0bd planemo upload for repository https://github.com/lldelisle/tools-lldelisle/tree/master/tools/fromgtfTobed12 commit 1aaffda5b95e0389e315179345642c0d005867c1
lldelisle
parents:
diff changeset
106 exons_starts = [e.start - 1
418e4d0fe0bd planemo upload for repository https://github.com/lldelisle/tools-lldelisle/tree/master/tools/fromgtfTobed12 commit 1aaffda5b95e0389e315179345642c0d005867c1
lldelisle
parents:
diff changeset
107 for e in
418e4d0fe0bd planemo upload for repository https://github.com/lldelisle/tools-lldelisle/tree/master/tools/fromgtfTobed12 commit 1aaffda5b95e0389e315179345642c0d005867c1
lldelisle
parents:
diff changeset
108 db.children(tr, featuretype='exon',
418e4d0fe0bd planemo upload for repository https://github.com/lldelisle/tools-lldelisle/tree/master/tools/fromgtfTobed12 commit 1aaffda5b95e0389e315179345642c0d005867c1
lldelisle
parents:
diff changeset
109 order_by='start')]
418e4d0fe0bd planemo upload for repository https://github.com/lldelisle/tools-lldelisle/tree/master/tools/fromgtfTobed12 commit 1aaffda5b95e0389e315179345642c0d005867c1
lldelisle
parents:
diff changeset
110 exons_length = [len(e)
418e4d0fe0bd planemo upload for repository https://github.com/lldelisle/tools-lldelisle/tree/master/tools/fromgtfTobed12 commit 1aaffda5b95e0389e315179345642c0d005867c1
lldelisle
parents:
diff changeset
111 for e in
418e4d0fe0bd planemo upload for repository https://github.com/lldelisle/tools-lldelisle/tree/master/tools/fromgtfTobed12 commit 1aaffda5b95e0389e315179345642c0d005867c1
lldelisle
parents:
diff changeset
112 db.children(tr, featuretype='exon',
418e4d0fe0bd planemo upload for repository https://github.com/lldelisle/tools-lldelisle/tree/master/tools/fromgtfTobed12 commit 1aaffda5b95e0389e315179345642c0d005867c1
lldelisle
parents:
diff changeset
113 order_by='start')]
418e4d0fe0bd planemo upload for repository https://github.com/lldelisle/tools-lldelisle/tree/master/tools/fromgtfTobed12 commit 1aaffda5b95e0389e315179345642c0d005867c1
lldelisle
parents:
diff changeset
114 # Rewrite the chromosome name if needed:
418e4d0fe0bd planemo upload for repository https://github.com/lldelisle/tools-lldelisle/tree/master/tools/fromgtfTobed12 commit 1aaffda5b95e0389e315179345642c0d005867c1
lldelisle
parents:
diff changeset
115 chrom = tr.chrom
418e4d0fe0bd planemo upload for repository https://github.com/lldelisle/tools-lldelisle/tree/master/tools/fromgtfTobed12 commit 1aaffda5b95e0389e315179345642c0d005867c1
lldelisle
parents:
diff changeset
116 if ucsc and chrom[0:3] != 'chr':
418e4d0fe0bd planemo upload for repository https://github.com/lldelisle/tools-lldelisle/tree/master/tools/fromgtfTobed12 commit 1aaffda5b95e0389e315179345642c0d005867c1
lldelisle
parents:
diff changeset
117 chrom = 'chr' + chrom
418e4d0fe0bd planemo upload for repository https://github.com/lldelisle/tools-lldelisle/tree/master/tools/fromgtfTobed12 commit 1aaffda5b95e0389e315179345642c0d005867c1
lldelisle
parents:
diff changeset
118 fo.write("%s\t%d\t%d\t%s\t%d\t%s\t%d\t%d\t%s\t%d\t%s\t%s\n" %
418e4d0fe0bd planemo upload for repository https://github.com/lldelisle/tools-lldelisle/tree/master/tools/fromgtfTobed12 commit 1aaffda5b95e0389e315179345642c0d005867c1
lldelisle
parents:
diff changeset
119 (chrom, tr.start - 1, tr.end, trName, 0, tr.strand,
418e4d0fe0bd planemo upload for repository https://github.com/lldelisle/tools-lldelisle/tree/master/tools/fromgtfTobed12 commit 1aaffda5b95e0389e315179345642c0d005867c1
lldelisle
parents:
diff changeset
120 cds_start, cds_end, "0", len(exons_starts),
418e4d0fe0bd planemo upload for repository https://github.com/lldelisle/tools-lldelisle/tree/master/tools/fromgtfTobed12 commit 1aaffda5b95e0389e315179345642c0d005867c1
lldelisle
parents:
diff changeset
121 ",".join([str(ex_l) for ex_l in exons_length]),
418e4d0fe0bd planemo upload for repository https://github.com/lldelisle/tools-lldelisle/tree/master/tools/fromgtfTobed12 commit 1aaffda5b95e0389e315179345642c0d005867c1
lldelisle
parents:
diff changeset
122 ",".join([str(s - (tr.start - 1)) for s in exons_starts])))
418e4d0fe0bd planemo upload for repository https://github.com/lldelisle/tools-lldelisle/tree/master/tools/fromgtfTobed12 commit 1aaffda5b95e0389e315179345642c0d005867c1
lldelisle
parents:
diff changeset
123
418e4d0fe0bd planemo upload for repository https://github.com/lldelisle/tools-lldelisle/tree/master/tools/fromgtfTobed12 commit 1aaffda5b95e0389e315179345642c0d005867c1
lldelisle
parents:
diff changeset
124
418e4d0fe0bd planemo upload for repository https://github.com/lldelisle/tools-lldelisle/tree/master/tools/fromgtfTobed12 commit 1aaffda5b95e0389e315179345642c0d005867c1
lldelisle
parents:
diff changeset
125 argp = argparse.ArgumentParser(
418e4d0fe0bd planemo upload for repository https://github.com/lldelisle/tools-lldelisle/tree/master/tools/fromgtfTobed12 commit 1aaffda5b95e0389e315179345642c0d005867c1
lldelisle
parents:
diff changeset
126 description=("Convert a gtf to a bed12 with one entry"
418e4d0fe0bd planemo upload for repository https://github.com/lldelisle/tools-lldelisle/tree/master/tools/fromgtfTobed12 commit 1aaffda5b95e0389e315179345642c0d005867c1
lldelisle
parents:
diff changeset
127 " per transcript/gene"))
418e4d0fe0bd planemo upload for repository https://github.com/lldelisle/tools-lldelisle/tree/master/tools/fromgtfTobed12 commit 1aaffda5b95e0389e315179345642c0d005867c1
lldelisle
parents:
diff changeset
128 argp.add_argument('input', default=None,
418e4d0fe0bd planemo upload for repository https://github.com/lldelisle/tools-lldelisle/tree/master/tools/fromgtfTobed12 commit 1aaffda5b95e0389e315179345642c0d005867c1
lldelisle
parents:
diff changeset
129 help="Input gtf file (can be gzip).")
418e4d0fe0bd planemo upload for repository https://github.com/lldelisle/tools-lldelisle/tree/master/tools/fromgtfTobed12 commit 1aaffda5b95e0389e315179345642c0d005867c1
lldelisle
parents:
diff changeset
130 argp.add_argument('--output', default=sys.stdout,
418e4d0fe0bd planemo upload for repository https://github.com/lldelisle/tools-lldelisle/tree/master/tools/fromgtfTobed12 commit 1aaffda5b95e0389e315179345642c0d005867c1
lldelisle
parents:
diff changeset
131 type=argparse.FileType('w'),
418e4d0fe0bd planemo upload for repository https://github.com/lldelisle/tools-lldelisle/tree/master/tools/fromgtfTobed12 commit 1aaffda5b95e0389e315179345642c0d005867c1
lldelisle
parents:
diff changeset
132 help="Output bed12 file.")
418e4d0fe0bd planemo upload for repository https://github.com/lldelisle/tools-lldelisle/tree/master/tools/fromgtfTobed12 commit 1aaffda5b95e0389e315179345642c0d005867c1
lldelisle
parents:
diff changeset
133 argp.add_argument('--ucscformat', action="store_true",
418e4d0fe0bd planemo upload for repository https://github.com/lldelisle/tools-lldelisle/tree/master/tools/fromgtfTobed12 commit 1aaffda5b95e0389e315179345642c0d005867c1
lldelisle
parents:
diff changeset
134 help="If you want that all chromosome names "
418e4d0fe0bd planemo upload for repository https://github.com/lldelisle/tools-lldelisle/tree/master/tools/fromgtfTobed12 commit 1aaffda5b95e0389e315179345642c0d005867c1
lldelisle
parents:
diff changeset
135 "begin with 'chr'.")
1
6fd4b3b90220 planemo upload for repository https://github.com/lldelisle/tools-lldelisle/tree/master/tools/fromgtfTobed12 commit 15b8c2cc83708044413a152322bcbfca8a74d29a
lldelisle
parents: 0
diff changeset
136 argp.add_argument('--preferedName', default=None,
6fd4b3b90220 planemo upload for repository https://github.com/lldelisle/tools-lldelisle/tree/master/tools/fromgtfTobed12 commit 15b8c2cc83708044413a152322bcbfca8a74d29a
lldelisle
parents: 0
diff changeset
137 help="Name to use for bed output.")
0
418e4d0fe0bd planemo upload for repository https://github.com/lldelisle/tools-lldelisle/tree/master/tools/fromgtfTobed12 commit 1aaffda5b95e0389e315179345642c0d005867c1
lldelisle
parents:
diff changeset
138 group = argp.add_mutually_exclusive_group()
418e4d0fe0bd planemo upload for repository https://github.com/lldelisle/tools-lldelisle/tree/master/tools/fromgtfTobed12 commit 1aaffda5b95e0389e315179345642c0d005867c1
lldelisle
parents:
diff changeset
139 group.add_argument('--mergeTranscripts', action="store_true",
418e4d0fe0bd planemo upload for repository https://github.com/lldelisle/tools-lldelisle/tree/master/tools/fromgtfTobed12 commit 1aaffda5b95e0389e315179345642c0d005867c1
lldelisle
parents:
diff changeset
140 help="Merge all transcripts into a single "
418e4d0fe0bd planemo upload for repository https://github.com/lldelisle/tools-lldelisle/tree/master/tools/fromgtfTobed12 commit 1aaffda5b95e0389e315179345642c0d005867c1
lldelisle
parents:
diff changeset
141 "entry to have one line per gene.")
418e4d0fe0bd planemo upload for repository https://github.com/lldelisle/tools-lldelisle/tree/master/tools/fromgtfTobed12 commit 1aaffda5b95e0389e315179345642c0d005867c1
lldelisle
parents:
diff changeset
142 group.add_argument('--mergeTranscriptsAndOverlappingExons',
418e4d0fe0bd planemo upload for repository https://github.com/lldelisle/tools-lldelisle/tree/master/tools/fromgtfTobed12 commit 1aaffda5b95e0389e315179345642c0d005867c1
lldelisle
parents:
diff changeset
143 action="store_true",
418e4d0fe0bd planemo upload for repository https://github.com/lldelisle/tools-lldelisle/tree/master/tools/fromgtfTobed12 commit 1aaffda5b95e0389e315179345642c0d005867c1
lldelisle
parents:
diff changeset
144 help="Merge all transcripts into a single "
418e4d0fe0bd planemo upload for repository https://github.com/lldelisle/tools-lldelisle/tree/master/tools/fromgtfTobed12 commit 1aaffda5b95e0389e315179345642c0d005867c1
lldelisle
parents:
diff changeset
145 "entry to have one line per gene and merge"
418e4d0fe0bd planemo upload for repository https://github.com/lldelisle/tools-lldelisle/tree/master/tools/fromgtfTobed12 commit 1aaffda5b95e0389e315179345642c0d005867c1
lldelisle
parents:
diff changeset
146 " overlapping exons.")
418e4d0fe0bd planemo upload for repository https://github.com/lldelisle/tools-lldelisle/tree/master/tools/fromgtfTobed12 commit 1aaffda5b95e0389e315179345642c0d005867c1
lldelisle
parents:
diff changeset
147
418e4d0fe0bd planemo upload for repository https://github.com/lldelisle/tools-lldelisle/tree/master/tools/fromgtfTobed12 commit 1aaffda5b95e0389e315179345642c0d005867c1
lldelisle
parents:
diff changeset
148 args = argp.parse_args()
1
6fd4b3b90220 planemo upload for repository https://github.com/lldelisle/tools-lldelisle/tree/master/tools/fromgtfTobed12 commit 15b8c2cc83708044413a152322bcbfca8a74d29a
lldelisle
parents: 0
diff changeset
149 convert_gtf_to_bed(args.input, args.output, args.preferedName,
0
418e4d0fe0bd planemo upload for repository https://github.com/lldelisle/tools-lldelisle/tree/master/tools/fromgtfTobed12 commit 1aaffda5b95e0389e315179345642c0d005867c1
lldelisle
parents:
diff changeset
150 args.mergeTranscripts,
418e4d0fe0bd planemo upload for repository https://github.com/lldelisle/tools-lldelisle/tree/master/tools/fromgtfTobed12 commit 1aaffda5b95e0389e315179345642c0d005867c1
lldelisle
parents:
diff changeset
151 args.mergeTranscriptsAndOverlappingExons,
418e4d0fe0bd planemo upload for repository https://github.com/lldelisle/tools-lldelisle/tree/master/tools/fromgtfTobed12 commit 1aaffda5b95e0389e315179345642c0d005867c1
lldelisle
parents:
diff changeset
152 args.ucscformat)