Mercurial > repos > galaxyp > translate_bed_sequences
annotate translate_bed_sequences.py @ 2:4221664a2bd0 draft default tip
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/translate_bed_sequences commit 2a470e2c775a7427aa530e058510e4dc7b6d8e80"
author | galaxyp |
---|---|
date | Tue, 07 Apr 2020 11:45:53 -0400 |
parents | d723eb657f1d |
children |
rev | line source |
---|---|
0
d723eb657f1d
planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/translate_bed_sequences commit e04ed4b4960d6109a85c1cc68a2bf4931c8751ef-dirty
galaxyp
parents:
diff
changeset
|
1 #!/usr/bin/env python |
d723eb657f1d
planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/translate_bed_sequences commit e04ed4b4960d6109a85c1cc68a2bf4931c8751ef-dirty
galaxyp
parents:
diff
changeset
|
2 """ |
d723eb657f1d
planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/translate_bed_sequences commit e04ed4b4960d6109a85c1cc68a2bf4931c8751ef-dirty
galaxyp
parents:
diff
changeset
|
3 # |
d723eb657f1d
planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/translate_bed_sequences commit e04ed4b4960d6109a85c1cc68a2bf4931c8751ef-dirty
galaxyp
parents:
diff
changeset
|
4 #------------------------------------------------------------------------------ |
d723eb657f1d
planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/translate_bed_sequences commit e04ed4b4960d6109a85c1cc68a2bf4931c8751ef-dirty
galaxyp
parents:
diff
changeset
|
5 # University of Minnesota |
d723eb657f1d
planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/translate_bed_sequences commit e04ed4b4960d6109a85c1cc68a2bf4931c8751ef-dirty
galaxyp
parents:
diff
changeset
|
6 # Copyright 2014, Regents of the University of Minnesota |
d723eb657f1d
planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/translate_bed_sequences commit e04ed4b4960d6109a85c1cc68a2bf4931c8751ef-dirty
galaxyp
parents:
diff
changeset
|
7 #------------------------------------------------------------------------------ |
d723eb657f1d
planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/translate_bed_sequences commit e04ed4b4960d6109a85c1cc68a2bf4931c8751ef-dirty
galaxyp
parents:
diff
changeset
|
8 # Author: |
d723eb657f1d
planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/translate_bed_sequences commit e04ed4b4960d6109a85c1cc68a2bf4931c8751ef-dirty
galaxyp
parents:
diff
changeset
|
9 # |
d723eb657f1d
planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/translate_bed_sequences commit e04ed4b4960d6109a85c1cc68a2bf4931c8751ef-dirty
galaxyp
parents:
diff
changeset
|
10 # James E Johnson |
d723eb657f1d
planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/translate_bed_sequences commit e04ed4b4960d6109a85c1cc68a2bf4931c8751ef-dirty
galaxyp
parents:
diff
changeset
|
11 # |
d723eb657f1d
planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/translate_bed_sequences commit e04ed4b4960d6109a85c1cc68a2bf4931c8751ef-dirty
galaxyp
parents:
diff
changeset
|
12 #------------------------------------------------------------------------------ |
d723eb657f1d
planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/translate_bed_sequences commit e04ed4b4960d6109a85c1cc68a2bf4931c8751ef-dirty
galaxyp
parents:
diff
changeset
|
13 |
d723eb657f1d
planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/translate_bed_sequences commit e04ed4b4960d6109a85c1cc68a2bf4931c8751ef-dirty
galaxyp
parents:
diff
changeset
|
14 Input: BED file (12 column) + 13th sequence column appended by extract_genomic_dna |
d723eb657f1d
planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/translate_bed_sequences commit e04ed4b4960d6109a85c1cc68a2bf4931c8751ef-dirty
galaxyp
parents:
diff
changeset
|
15 Output: Fasta of 3-frame translations of the spliced sequence |
d723eb657f1d
planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/translate_bed_sequences commit e04ed4b4960d6109a85c1cc68a2bf4931c8751ef-dirty
galaxyp
parents:
diff
changeset
|
16 """ |
d723eb657f1d
planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/translate_bed_sequences commit e04ed4b4960d6109a85c1cc68a2bf4931c8751ef-dirty
galaxyp
parents:
diff
changeset
|
17 |
2
4221664a2bd0
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/translate_bed_sequences commit 2a470e2c775a7427aa530e058510e4dc7b6d8e80"
galaxyp
parents:
0
diff
changeset
|
18 import optparse |
4221664a2bd0
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/translate_bed_sequences commit 2a470e2c775a7427aa530e058510e4dc7b6d8e80"
galaxyp
parents:
0
diff
changeset
|
19 import os.path |
4221664a2bd0
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/translate_bed_sequences commit 2a470e2c775a7427aa530e058510e4dc7b6d8e80"
galaxyp
parents:
0
diff
changeset
|
20 import sys |
0
d723eb657f1d
planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/translate_bed_sequences commit e04ed4b4960d6109a85c1cc68a2bf4931c8751ef-dirty
galaxyp
parents:
diff
changeset
|
21 import tempfile |
2
4221664a2bd0
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/translate_bed_sequences commit 2a470e2c775a7427aa530e058510e4dc7b6d8e80"
galaxyp
parents:
0
diff
changeset
|
22 |
4221664a2bd0
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/translate_bed_sequences commit 2a470e2c775a7427aa530e058510e4dc7b6d8e80"
galaxyp
parents:
0
diff
changeset
|
23 from Bio.Seq import ( |
4221664a2bd0
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/translate_bed_sequences commit 2a470e2c775a7427aa530e058510e4dc7b6d8e80"
galaxyp
parents:
0
diff
changeset
|
24 reverse_complement, |
4221664a2bd0
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/translate_bed_sequences commit 2a470e2c775a7427aa530e058510e4dc7b6d8e80"
galaxyp
parents:
0
diff
changeset
|
25 translate |
4221664a2bd0
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/translate_bed_sequences commit 2a470e2c775a7427aa530e058510e4dc7b6d8e80"
galaxyp
parents:
0
diff
changeset
|
26 ) |
4221664a2bd0
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/translate_bed_sequences commit 2a470e2c775a7427aa530e058510e4dc7b6d8e80"
galaxyp
parents:
0
diff
changeset
|
27 |
4221664a2bd0
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/translate_bed_sequences commit 2a470e2c775a7427aa530e058510e4dc7b6d8e80"
galaxyp
parents:
0
diff
changeset
|
28 |
4221664a2bd0
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/translate_bed_sequences commit 2a470e2c775a7427aa530e058510e4dc7b6d8e80"
galaxyp
parents:
0
diff
changeset
|
29 class BedEntry(object): |
4221664a2bd0
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/translate_bed_sequences commit 2a470e2c775a7427aa530e058510e4dc7b6d8e80"
galaxyp
parents:
0
diff
changeset
|
30 def __init__(self, line): |
4221664a2bd0
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/translate_bed_sequences commit 2a470e2c775a7427aa530e058510e4dc7b6d8e80"
galaxyp
parents:
0
diff
changeset
|
31 self.line = line |
4221664a2bd0
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/translate_bed_sequences commit 2a470e2c775a7427aa530e058510e4dc7b6d8e80"
galaxyp
parents:
0
diff
changeset
|
32 try: |
4221664a2bd0
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/translate_bed_sequences commit 2a470e2c775a7427aa530e058510e4dc7b6d8e80"
galaxyp
parents:
0
diff
changeset
|
33 fields = line.rstrip('\r\n').split('\t') |
4221664a2bd0
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/translate_bed_sequences commit 2a470e2c775a7427aa530e058510e4dc7b6d8e80"
galaxyp
parents:
0
diff
changeset
|
34 (chrom, chromStart, chromEnd, name, score, strand, thickStart, thickEnd, itemRgb, blockCount, blockSizes, blockStarts) = fields[0:12] |
4221664a2bd0
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/translate_bed_sequences commit 2a470e2c775a7427aa530e058510e4dc7b6d8e80"
galaxyp
parents:
0
diff
changeset
|
35 seq = fields[12] if len(fields) > 12 else None |
4221664a2bd0
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/translate_bed_sequences commit 2a470e2c775a7427aa530e058510e4dc7b6d8e80"
galaxyp
parents:
0
diff
changeset
|
36 self.chrom = chrom |
4221664a2bd0
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/translate_bed_sequences commit 2a470e2c775a7427aa530e058510e4dc7b6d8e80"
galaxyp
parents:
0
diff
changeset
|
37 self.chromStart = int(chromStart) |
4221664a2bd0
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/translate_bed_sequences commit 2a470e2c775a7427aa530e058510e4dc7b6d8e80"
galaxyp
parents:
0
diff
changeset
|
38 self.chromEnd = int(chromEnd) |
4221664a2bd0
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/translate_bed_sequences commit 2a470e2c775a7427aa530e058510e4dc7b6d8e80"
galaxyp
parents:
0
diff
changeset
|
39 self.name = name |
4221664a2bd0
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/translate_bed_sequences commit 2a470e2c775a7427aa530e058510e4dc7b6d8e80"
galaxyp
parents:
0
diff
changeset
|
40 self.score = int(score) |
4221664a2bd0
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/translate_bed_sequences commit 2a470e2c775a7427aa530e058510e4dc7b6d8e80"
galaxyp
parents:
0
diff
changeset
|
41 self.strand = strand |
4221664a2bd0
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/translate_bed_sequences commit 2a470e2c775a7427aa530e058510e4dc7b6d8e80"
galaxyp
parents:
0
diff
changeset
|
42 self.thickStart = int(thickStart) |
4221664a2bd0
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/translate_bed_sequences commit 2a470e2c775a7427aa530e058510e4dc7b6d8e80"
galaxyp
parents:
0
diff
changeset
|
43 self.thickEnd = int(thickEnd) |
4221664a2bd0
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/translate_bed_sequences commit 2a470e2c775a7427aa530e058510e4dc7b6d8e80"
galaxyp
parents:
0
diff
changeset
|
44 self.itemRgb = itemRgb |
4221664a2bd0
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/translate_bed_sequences commit 2a470e2c775a7427aa530e058510e4dc7b6d8e80"
galaxyp
parents:
0
diff
changeset
|
45 self.blockCount = int(blockCount) |
4221664a2bd0
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/translate_bed_sequences commit 2a470e2c775a7427aa530e058510e4dc7b6d8e80"
galaxyp
parents:
0
diff
changeset
|
46 self.blockSizes = [int(x) for x in blockSizes.split(',')] |
4221664a2bd0
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/translate_bed_sequences commit 2a470e2c775a7427aa530e058510e4dc7b6d8e80"
galaxyp
parents:
0
diff
changeset
|
47 self.blockStarts = [int(x) for x in blockStarts.split(',')] |
4221664a2bd0
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/translate_bed_sequences commit 2a470e2c775a7427aa530e058510e4dc7b6d8e80"
galaxyp
parents:
0
diff
changeset
|
48 self.seq = seq |
4221664a2bd0
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/translate_bed_sequences commit 2a470e2c775a7427aa530e058510e4dc7b6d8e80"
galaxyp
parents:
0
diff
changeset
|
49 except Exception as e: |
4221664a2bd0
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/translate_bed_sequences commit 2a470e2c775a7427aa530e058510e4dc7b6d8e80"
galaxyp
parents:
0
diff
changeset
|
50 sys.stderr.write("Unable to read Bed entry %s\n" % e) |
4221664a2bd0
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/translate_bed_sequences commit 2a470e2c775a7427aa530e058510e4dc7b6d8e80"
galaxyp
parents:
0
diff
changeset
|
51 exit(1) |
4221664a2bd0
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/translate_bed_sequences commit 2a470e2c775a7427aa530e058510e4dc7b6d8e80"
galaxyp
parents:
0
diff
changeset
|
52 |
4221664a2bd0
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/translate_bed_sequences commit 2a470e2c775a7427aa530e058510e4dc7b6d8e80"
galaxyp
parents:
0
diff
changeset
|
53 def __str__(self): |
4221664a2bd0
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/translate_bed_sequences commit 2a470e2c775a7427aa530e058510e4dc7b6d8e80"
galaxyp
parents:
0
diff
changeset
|
54 return '%s\t%d\t%d\t%s\t%d\t%s\t%d\t%d\t%s\t%d\t%s\t%s%s' % ( |
4221664a2bd0
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/translate_bed_sequences commit 2a470e2c775a7427aa530e058510e4dc7b6d8e80"
galaxyp
parents:
0
diff
changeset
|
55 self.chrom, self.chromStart, self.chromEnd, self.name, self.score, self.strand, self.thickStart, self.thickEnd, self.itemRgb, self.blockCount, |
4221664a2bd0
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/translate_bed_sequences commit 2a470e2c775a7427aa530e058510e4dc7b6d8e80"
galaxyp
parents:
0
diff
changeset
|
56 ','.join([str(x) for x in self.blockSizes]), |
4221664a2bd0
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/translate_bed_sequences commit 2a470e2c775a7427aa530e058510e4dc7b6d8e80"
galaxyp
parents:
0
diff
changeset
|
57 ','.join([str(x) for x in self.blockStarts]), |
4221664a2bd0
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/translate_bed_sequences commit 2a470e2c775a7427aa530e058510e4dc7b6d8e80"
galaxyp
parents:
0
diff
changeset
|
58 '\t%s' % self.seq if self.seq else '') |
4221664a2bd0
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/translate_bed_sequences commit 2a470e2c775a7427aa530e058510e4dc7b6d8e80"
galaxyp
parents:
0
diff
changeset
|
59 |
4221664a2bd0
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/translate_bed_sequences commit 2a470e2c775a7427aa530e058510e4dc7b6d8e80"
galaxyp
parents:
0
diff
changeset
|
60 def get_splice_junctions(self): |
4221664a2bd0
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/translate_bed_sequences commit 2a470e2c775a7427aa530e058510e4dc7b6d8e80"
galaxyp
parents:
0
diff
changeset
|
61 splice_juncs = [] |
4221664a2bd0
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/translate_bed_sequences commit 2a470e2c775a7427aa530e058510e4dc7b6d8e80"
galaxyp
parents:
0
diff
changeset
|
62 for i in range(self.blockCount - 1): |
4221664a2bd0
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/translate_bed_sequences commit 2a470e2c775a7427aa530e058510e4dc7b6d8e80"
galaxyp
parents:
0
diff
changeset
|
63 splice_junc = "%s:%d_%d" % (self.chrom, self.chromStart + self.blockSizes[i], self.chromStart + self.blockStarts[i + 1]) |
4221664a2bd0
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/translate_bed_sequences commit 2a470e2c775a7427aa530e058510e4dc7b6d8e80"
galaxyp
parents:
0
diff
changeset
|
64 splice_juncs.append(splice_junc) |
4221664a2bd0
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/translate_bed_sequences commit 2a470e2c775a7427aa530e058510e4dc7b6d8e80"
galaxyp
parents:
0
diff
changeset
|
65 return splice_juncs |
4221664a2bd0
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/translate_bed_sequences commit 2a470e2c775a7427aa530e058510e4dc7b6d8e80"
galaxyp
parents:
0
diff
changeset
|
66 |
4221664a2bd0
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/translate_bed_sequences commit 2a470e2c775a7427aa530e058510e4dc7b6d8e80"
galaxyp
parents:
0
diff
changeset
|
67 def get_exon_seqs(self): |
4221664a2bd0
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/translate_bed_sequences commit 2a470e2c775a7427aa530e058510e4dc7b6d8e80"
galaxyp
parents:
0
diff
changeset
|
68 exons = [] |
4221664a2bd0
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/translate_bed_sequences commit 2a470e2c775a7427aa530e058510e4dc7b6d8e80"
galaxyp
parents:
0
diff
changeset
|
69 for i in range(self.blockCount): |
4221664a2bd0
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/translate_bed_sequences commit 2a470e2c775a7427aa530e058510e4dc7b6d8e80"
galaxyp
parents:
0
diff
changeset
|
70 # splice_junc = "%s:%d_%d" % (self.chrom, self.chromStart + self.blockSizes[i], self.chromStart + self.blockStarts[i+1]) |
4221664a2bd0
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/translate_bed_sequences commit 2a470e2c775a7427aa530e058510e4dc7b6d8e80"
galaxyp
parents:
0
diff
changeset
|
71 exons.append(self.seq[self.blockStarts[i]:self.blockStarts[i] + self.blockSizes[i]]) |
4221664a2bd0
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/translate_bed_sequences commit 2a470e2c775a7427aa530e058510e4dc7b6d8e80"
galaxyp
parents:
0
diff
changeset
|
72 if self.strand == '-': # reverse complement |
4221664a2bd0
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/translate_bed_sequences commit 2a470e2c775a7427aa530e058510e4dc7b6d8e80"
galaxyp
parents:
0
diff
changeset
|
73 exons.reverse() |
4221664a2bd0
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/translate_bed_sequences commit 2a470e2c775a7427aa530e058510e4dc7b6d8e80"
galaxyp
parents:
0
diff
changeset
|
74 for i, s in enumerate(exons): |
4221664a2bd0
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/translate_bed_sequences commit 2a470e2c775a7427aa530e058510e4dc7b6d8e80"
galaxyp
parents:
0
diff
changeset
|
75 exons[i] = reverse_complement(s) |
4221664a2bd0
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/translate_bed_sequences commit 2a470e2c775a7427aa530e058510e4dc7b6d8e80"
galaxyp
parents:
0
diff
changeset
|
76 return exons |
4221664a2bd0
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/translate_bed_sequences commit 2a470e2c775a7427aa530e058510e4dc7b6d8e80"
galaxyp
parents:
0
diff
changeset
|
77 |
4221664a2bd0
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/translate_bed_sequences commit 2a470e2c775a7427aa530e058510e4dc7b6d8e80"
galaxyp
parents:
0
diff
changeset
|
78 def get_spliced_seq(self): |
4221664a2bd0
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/translate_bed_sequences commit 2a470e2c775a7427aa530e058510e4dc7b6d8e80"
galaxyp
parents:
0
diff
changeset
|
79 seq = ''.join(self.get_exon_seqs()) |
4221664a2bd0
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/translate_bed_sequences commit 2a470e2c775a7427aa530e058510e4dc7b6d8e80"
galaxyp
parents:
0
diff
changeset
|
80 return seq |
4221664a2bd0
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/translate_bed_sequences commit 2a470e2c775a7427aa530e058510e4dc7b6d8e80"
galaxyp
parents:
0
diff
changeset
|
81 |
4221664a2bd0
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/translate_bed_sequences commit 2a470e2c775a7427aa530e058510e4dc7b6d8e80"
galaxyp
parents:
0
diff
changeset
|
82 def get_translation(self, sequence=None): |
4221664a2bd0
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/translate_bed_sequences commit 2a470e2c775a7427aa530e058510e4dc7b6d8e80"
galaxyp
parents:
0
diff
changeset
|
83 translation = None |
4221664a2bd0
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/translate_bed_sequences commit 2a470e2c775a7427aa530e058510e4dc7b6d8e80"
galaxyp
parents:
0
diff
changeset
|
84 seq = sequence if sequence else self.get_spliced_seq() |
4221664a2bd0
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/translate_bed_sequences commit 2a470e2c775a7427aa530e058510e4dc7b6d8e80"
galaxyp
parents:
0
diff
changeset
|
85 if seq: |
4221664a2bd0
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/translate_bed_sequences commit 2a470e2c775a7427aa530e058510e4dc7b6d8e80"
galaxyp
parents:
0
diff
changeset
|
86 seqlen = int(len(seq) / 3) * 3 |
4221664a2bd0
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/translate_bed_sequences commit 2a470e2c775a7427aa530e058510e4dc7b6d8e80"
galaxyp
parents:
0
diff
changeset
|
87 if seqlen >= 3: |
4221664a2bd0
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/translate_bed_sequences commit 2a470e2c775a7427aa530e058510e4dc7b6d8e80"
galaxyp
parents:
0
diff
changeset
|
88 translation = translate(seq[:seqlen]) |
4221664a2bd0
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/translate_bed_sequences commit 2a470e2c775a7427aa530e058510e4dc7b6d8e80"
galaxyp
parents:
0
diff
changeset
|
89 return translation |
4221664a2bd0
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/translate_bed_sequences commit 2a470e2c775a7427aa530e058510e4dc7b6d8e80"
galaxyp
parents:
0
diff
changeset
|
90 |
4221664a2bd0
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/translate_bed_sequences commit 2a470e2c775a7427aa530e058510e4dc7b6d8e80"
galaxyp
parents:
0
diff
changeset
|
91 def get_translations(self): |
4221664a2bd0
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/translate_bed_sequences commit 2a470e2c775a7427aa530e058510e4dc7b6d8e80"
galaxyp
parents:
0
diff
changeset
|
92 translations = [] |
4221664a2bd0
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/translate_bed_sequences commit 2a470e2c775a7427aa530e058510e4dc7b6d8e80"
galaxyp
parents:
0
diff
changeset
|
93 seq = self.get_spliced_seq() |
4221664a2bd0
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/translate_bed_sequences commit 2a470e2c775a7427aa530e058510e4dc7b6d8e80"
galaxyp
parents:
0
diff
changeset
|
94 if seq: |
4221664a2bd0
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/translate_bed_sequences commit 2a470e2c775a7427aa530e058510e4dc7b6d8e80"
galaxyp
parents:
0
diff
changeset
|
95 for i in range(3): |
4221664a2bd0
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/translate_bed_sequences commit 2a470e2c775a7427aa530e058510e4dc7b6d8e80"
galaxyp
parents:
0
diff
changeset
|
96 translation = self.get_translation(sequence=seq[i:]) |
4221664a2bd0
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/translate_bed_sequences commit 2a470e2c775a7427aa530e058510e4dc7b6d8e80"
galaxyp
parents:
0
diff
changeset
|
97 if translation: |
4221664a2bd0
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/translate_bed_sequences commit 2a470e2c775a7427aa530e058510e4dc7b6d8e80"
galaxyp
parents:
0
diff
changeset
|
98 translations.append(translation) |
4221664a2bd0
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/translate_bed_sequences commit 2a470e2c775a7427aa530e058510e4dc7b6d8e80"
galaxyp
parents:
0
diff
changeset
|
99 return translations |
4221664a2bd0
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/translate_bed_sequences commit 2a470e2c775a7427aa530e058510e4dc7b6d8e80"
galaxyp
parents:
0
diff
changeset
|
100 |
4221664a2bd0
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/translate_bed_sequences commit 2a470e2c775a7427aa530e058510e4dc7b6d8e80"
galaxyp
parents:
0
diff
changeset
|
101 def get_subrange(self, tstart, tstop): |
4221664a2bd0
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/translate_bed_sequences commit 2a470e2c775a7427aa530e058510e4dc7b6d8e80"
galaxyp
parents:
0
diff
changeset
|
102 """ |
4221664a2bd0
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/translate_bed_sequences commit 2a470e2c775a7427aa530e058510e4dc7b6d8e80"
galaxyp
parents:
0
diff
changeset
|
103 (start, end) |
4221664a2bd0
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/translate_bed_sequences commit 2a470e2c775a7427aa530e058510e4dc7b6d8e80"
galaxyp
parents:
0
diff
changeset
|
104 """ |
4221664a2bd0
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/translate_bed_sequences commit 2a470e2c775a7427aa530e058510e4dc7b6d8e80"
galaxyp
parents:
0
diff
changeset
|
105 chromStart = self.chromStart |
4221664a2bd0
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/translate_bed_sequences commit 2a470e2c775a7427aa530e058510e4dc7b6d8e80"
galaxyp
parents:
0
diff
changeset
|
106 chromEnd = self.chromEnd |
4221664a2bd0
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/translate_bed_sequences commit 2a470e2c775a7427aa530e058510e4dc7b6d8e80"
galaxyp
parents:
0
diff
changeset
|
107 r = range(self.blockCount) |
4221664a2bd0
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/translate_bed_sequences commit 2a470e2c775a7427aa530e058510e4dc7b6d8e80"
galaxyp
parents:
0
diff
changeset
|
108 if self.strand == '-': |
4221664a2bd0
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/translate_bed_sequences commit 2a470e2c775a7427aa530e058510e4dc7b6d8e80"
galaxyp
parents:
0
diff
changeset
|
109 r = list(r) |
4221664a2bd0
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/translate_bed_sequences commit 2a470e2c775a7427aa530e058510e4dc7b6d8e80"
galaxyp
parents:
0
diff
changeset
|
110 r.reverse() |
4221664a2bd0
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/translate_bed_sequences commit 2a470e2c775a7427aa530e058510e4dc7b6d8e80"
galaxyp
parents:
0
diff
changeset
|
111 bStart = 0 |
4221664a2bd0
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/translate_bed_sequences commit 2a470e2c775a7427aa530e058510e4dc7b6d8e80"
galaxyp
parents:
0
diff
changeset
|
112 for x in r: |
4221664a2bd0
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/translate_bed_sequences commit 2a470e2c775a7427aa530e058510e4dc7b6d8e80"
galaxyp
parents:
0
diff
changeset
|
113 bEnd = bStart + self.blockSizes[x] |
4221664a2bd0
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/translate_bed_sequences commit 2a470e2c775a7427aa530e058510e4dc7b6d8e80"
galaxyp
parents:
0
diff
changeset
|
114 if bStart <= tstart < bEnd: |
4221664a2bd0
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/translate_bed_sequences commit 2a470e2c775a7427aa530e058510e4dc7b6d8e80"
galaxyp
parents:
0
diff
changeset
|
115 if self.strand == '+': |
4221664a2bd0
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/translate_bed_sequences commit 2a470e2c775a7427aa530e058510e4dc7b6d8e80"
galaxyp
parents:
0
diff
changeset
|
116 chromStart = self.chromStart + self.blockStarts[x] + (tstart - bStart) |
4221664a2bd0
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/translate_bed_sequences commit 2a470e2c775a7427aa530e058510e4dc7b6d8e80"
galaxyp
parents:
0
diff
changeset
|
117 else: |
4221664a2bd0
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/translate_bed_sequences commit 2a470e2c775a7427aa530e058510e4dc7b6d8e80"
galaxyp
parents:
0
diff
changeset
|
118 chromEnd = self.chromStart + self.blockStarts[x] + (tstart - bStart) |
4221664a2bd0
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/translate_bed_sequences commit 2a470e2c775a7427aa530e058510e4dc7b6d8e80"
galaxyp
parents:
0
diff
changeset
|
119 if bStart <= tstop < bEnd: |
4221664a2bd0
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/translate_bed_sequences commit 2a470e2c775a7427aa530e058510e4dc7b6d8e80"
galaxyp
parents:
0
diff
changeset
|
120 if self.strand == '+': |
4221664a2bd0
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/translate_bed_sequences commit 2a470e2c775a7427aa530e058510e4dc7b6d8e80"
galaxyp
parents:
0
diff
changeset
|
121 chromEnd = self.chromStart + self.blockStarts[x] + (tstop - bStart) |
4221664a2bd0
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/translate_bed_sequences commit 2a470e2c775a7427aa530e058510e4dc7b6d8e80"
galaxyp
parents:
0
diff
changeset
|
122 else: |
4221664a2bd0
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/translate_bed_sequences commit 2a470e2c775a7427aa530e058510e4dc7b6d8e80"
galaxyp
parents:
0
diff
changeset
|
123 chromStart = self.chromStart + self.blockStarts[x] + self.blockSizes[x] - (tstop - bStart) |
4221664a2bd0
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/translate_bed_sequences commit 2a470e2c775a7427aa530e058510e4dc7b6d8e80"
galaxyp
parents:
0
diff
changeset
|
124 bStart += self.blockSizes[x] |
4221664a2bd0
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/translate_bed_sequences commit 2a470e2c775a7427aa530e058510e4dc7b6d8e80"
galaxyp
parents:
0
diff
changeset
|
125 return(chromStart, chromEnd) |
0
d723eb657f1d
planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/translate_bed_sequences commit e04ed4b4960d6109a85c1cc68a2bf4931c8751ef-dirty
galaxyp
parents:
diff
changeset
|
126 |
2
4221664a2bd0
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/translate_bed_sequences commit 2a470e2c775a7427aa530e058510e4dc7b6d8e80"
galaxyp
parents:
0
diff
changeset
|
127 def get_blocks(self, chromStart, chromEnd): |
4221664a2bd0
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/translate_bed_sequences commit 2a470e2c775a7427aa530e058510e4dc7b6d8e80"
galaxyp
parents:
0
diff
changeset
|
128 """ |
4221664a2bd0
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/translate_bed_sequences commit 2a470e2c775a7427aa530e058510e4dc7b6d8e80"
galaxyp
parents:
0
diff
changeset
|
129 get the blocks for sub range |
4221664a2bd0
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/translate_bed_sequences commit 2a470e2c775a7427aa530e058510e4dc7b6d8e80"
galaxyp
parents:
0
diff
changeset
|
130 """ |
4221664a2bd0
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/translate_bed_sequences commit 2a470e2c775a7427aa530e058510e4dc7b6d8e80"
galaxyp
parents:
0
diff
changeset
|
131 tblockCount = 0 |
4221664a2bd0
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/translate_bed_sequences commit 2a470e2c775a7427aa530e058510e4dc7b6d8e80"
galaxyp
parents:
0
diff
changeset
|
132 tblockSizes = [] |
4221664a2bd0
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/translate_bed_sequences commit 2a470e2c775a7427aa530e058510e4dc7b6d8e80"
galaxyp
parents:
0
diff
changeset
|
133 tblockStarts = [] |
4221664a2bd0
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/translate_bed_sequences commit 2a470e2c775a7427aa530e058510e4dc7b6d8e80"
galaxyp
parents:
0
diff
changeset
|
134 for x in range(self.blockCount): |
4221664a2bd0
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/translate_bed_sequences commit 2a470e2c775a7427aa530e058510e4dc7b6d8e80"
galaxyp
parents:
0
diff
changeset
|
135 bStart = self.chromStart + self.blockStarts[x] |
4221664a2bd0
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/translate_bed_sequences commit 2a470e2c775a7427aa530e058510e4dc7b6d8e80"
galaxyp
parents:
0
diff
changeset
|
136 bEnd = bStart + self.blockSizes[x] |
4221664a2bd0
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/translate_bed_sequences commit 2a470e2c775a7427aa530e058510e4dc7b6d8e80"
galaxyp
parents:
0
diff
changeset
|
137 if bStart > chromEnd: |
4221664a2bd0
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/translate_bed_sequences commit 2a470e2c775a7427aa530e058510e4dc7b6d8e80"
galaxyp
parents:
0
diff
changeset
|
138 break |
4221664a2bd0
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/translate_bed_sequences commit 2a470e2c775a7427aa530e058510e4dc7b6d8e80"
galaxyp
parents:
0
diff
changeset
|
139 if bEnd < chromStart: |
4221664a2bd0
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/translate_bed_sequences commit 2a470e2c775a7427aa530e058510e4dc7b6d8e80"
galaxyp
parents:
0
diff
changeset
|
140 continue |
4221664a2bd0
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/translate_bed_sequences commit 2a470e2c775a7427aa530e058510e4dc7b6d8e80"
galaxyp
parents:
0
diff
changeset
|
141 cStart = max(chromStart, bStart) |
4221664a2bd0
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/translate_bed_sequences commit 2a470e2c775a7427aa530e058510e4dc7b6d8e80"
galaxyp
parents:
0
diff
changeset
|
142 tblockStarts.append(cStart - chromStart) |
4221664a2bd0
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/translate_bed_sequences commit 2a470e2c775a7427aa530e058510e4dc7b6d8e80"
galaxyp
parents:
0
diff
changeset
|
143 tblockSizes.append(min(chromEnd, bEnd) - cStart) |
4221664a2bd0
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/translate_bed_sequences commit 2a470e2c775a7427aa530e058510e4dc7b6d8e80"
galaxyp
parents:
0
diff
changeset
|
144 tblockCount += 1 |
4221664a2bd0
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/translate_bed_sequences commit 2a470e2c775a7427aa530e058510e4dc7b6d8e80"
galaxyp
parents:
0
diff
changeset
|
145 # print >> sys.stderr, "tblockCount: %d tblockStarts: %s tblockSizes: %s" % (tblockCount, tblockStarts, tblockSizes) |
4221664a2bd0
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/translate_bed_sequences commit 2a470e2c775a7427aa530e058510e4dc7b6d8e80"
galaxyp
parents:
0
diff
changeset
|
146 return (tblockCount, tblockSizes, tblockStarts) |
4221664a2bd0
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/translate_bed_sequences commit 2a470e2c775a7427aa530e058510e4dc7b6d8e80"
galaxyp
parents:
0
diff
changeset
|
147 |
4221664a2bd0
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/translate_bed_sequences commit 2a470e2c775a7427aa530e058510e4dc7b6d8e80"
galaxyp
parents:
0
diff
changeset
|
148 def get_filterd_translations(self, untrimmed=False, filtering=True, ignore_left_bp=0, ignore_right_bp=0, debug=False): |
4221664a2bd0
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/translate_bed_sequences commit 2a470e2c775a7427aa530e058510e4dc7b6d8e80"
galaxyp
parents:
0
diff
changeset
|
149 """ |
4221664a2bd0
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/translate_bed_sequences commit 2a470e2c775a7427aa530e058510e4dc7b6d8e80"
galaxyp
parents:
0
diff
changeset
|
150 [(start, end, seq, blockCount, blockSizes, blockStarts), (start, end, seq, blockCount, blockSizes, blockStarts), (start, end, seq, blockCount, blockSizes, blockStarts)] |
4221664a2bd0
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/translate_bed_sequences commit 2a470e2c775a7427aa530e058510e4dc7b6d8e80"
galaxyp
parents:
0
diff
changeset
|
151 filter: ignore translation if stop codon in first exon after ignore_left_bp |
4221664a2bd0
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/translate_bed_sequences commit 2a470e2c775a7427aa530e058510e4dc7b6d8e80"
galaxyp
parents:
0
diff
changeset
|
152 """ |
4221664a2bd0
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/translate_bed_sequences commit 2a470e2c775a7427aa530e058510e4dc7b6d8e80"
galaxyp
parents:
0
diff
changeset
|
153 translations = [None, None, None, None, None, None] |
4221664a2bd0
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/translate_bed_sequences commit 2a470e2c775a7427aa530e058510e4dc7b6d8e80"
galaxyp
parents:
0
diff
changeset
|
154 seq = self.get_spliced_seq() |
4221664a2bd0
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/translate_bed_sequences commit 2a470e2c775a7427aa530e058510e4dc7b6d8e80"
galaxyp
parents:
0
diff
changeset
|
155 ignore = int((ignore_left_bp if self.strand == '+' else ignore_right_bp) / 3) |
4221664a2bd0
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/translate_bed_sequences commit 2a470e2c775a7427aa530e058510e4dc7b6d8e80"
galaxyp
parents:
0
diff
changeset
|
156 block_sum = sum(self.blockSizes) |
4221664a2bd0
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/translate_bed_sequences commit 2a470e2c775a7427aa530e058510e4dc7b6d8e80"
galaxyp
parents:
0
diff
changeset
|
157 exon_sizes = [x for x in self.blockSizes] |
4221664a2bd0
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/translate_bed_sequences commit 2a470e2c775a7427aa530e058510e4dc7b6d8e80"
galaxyp
parents:
0
diff
changeset
|
158 if self.strand == '-': |
4221664a2bd0
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/translate_bed_sequences commit 2a470e2c775a7427aa530e058510e4dc7b6d8e80"
galaxyp
parents:
0
diff
changeset
|
159 exon_sizes.reverse() |
4221664a2bd0
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/translate_bed_sequences commit 2a470e2c775a7427aa530e058510e4dc7b6d8e80"
galaxyp
parents:
0
diff
changeset
|
160 splice_sites = [int(sum(exon_sizes[:x]) / 3) for x in range(1, len(exon_sizes))] |
4221664a2bd0
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/translate_bed_sequences commit 2a470e2c775a7427aa530e058510e4dc7b6d8e80"
galaxyp
parents:
0
diff
changeset
|
161 if debug: |
4221664a2bd0
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/translate_bed_sequences commit 2a470e2c775a7427aa530e058510e4dc7b6d8e80"
galaxyp
parents:
0
diff
changeset
|
162 sys.stderr.write("splice_sites: %s\n" % splice_sites) |
4221664a2bd0
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/translate_bed_sequences commit 2a470e2c775a7427aa530e058510e4dc7b6d8e80"
galaxyp
parents:
0
diff
changeset
|
163 junc = splice_sites[0] if len(splice_sites) > 0 else exon_sizes[0] |
4221664a2bd0
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/translate_bed_sequences commit 2a470e2c775a7427aa530e058510e4dc7b6d8e80"
galaxyp
parents:
0
diff
changeset
|
164 if seq: |
4221664a2bd0
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/translate_bed_sequences commit 2a470e2c775a7427aa530e058510e4dc7b6d8e80"
galaxyp
parents:
0
diff
changeset
|
165 for i in range(3): |
4221664a2bd0
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/translate_bed_sequences commit 2a470e2c775a7427aa530e058510e4dc7b6d8e80"
galaxyp
parents:
0
diff
changeset
|
166 translation = self.get_translation(sequence=seq[i:]) |
4221664a2bd0
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/translate_bed_sequences commit 2a470e2c775a7427aa530e058510e4dc7b6d8e80"
galaxyp
parents:
0
diff
changeset
|
167 if translation: |
4221664a2bd0
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/translate_bed_sequences commit 2a470e2c775a7427aa530e058510e4dc7b6d8e80"
galaxyp
parents:
0
diff
changeset
|
168 tstart = 0 |
4221664a2bd0
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/translate_bed_sequences commit 2a470e2c775a7427aa530e058510e4dc7b6d8e80"
galaxyp
parents:
0
diff
changeset
|
169 tstop = len(translation) |
4221664a2bd0
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/translate_bed_sequences commit 2a470e2c775a7427aa530e058510e4dc7b6d8e80"
galaxyp
parents:
0
diff
changeset
|
170 offset = (block_sum - i) % 3 |
4221664a2bd0
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/translate_bed_sequences commit 2a470e2c775a7427aa530e058510e4dc7b6d8e80"
galaxyp
parents:
0
diff
changeset
|
171 if debug: |
4221664a2bd0
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/translate_bed_sequences commit 2a470e2c775a7427aa530e058510e4dc7b6d8e80"
galaxyp
parents:
0
diff
changeset
|
172 sys.stderr.write("frame: %d\ttstart: %d tstop: %d offset: %d\t%s\n" % (i, tstart, tstop, offset, translation)) |
4221664a2bd0
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/translate_bed_sequences commit 2a470e2c775a7427aa530e058510e4dc7b6d8e80"
galaxyp
parents:
0
diff
changeset
|
173 if not untrimmed: |
4221664a2bd0
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/translate_bed_sequences commit 2a470e2c775a7427aa530e058510e4dc7b6d8e80"
galaxyp
parents:
0
diff
changeset
|
174 tstart = translation.rfind('*', 0, junc) + 1 |
4221664a2bd0
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/translate_bed_sequences commit 2a470e2c775a7427aa530e058510e4dc7b6d8e80"
galaxyp
parents:
0
diff
changeset
|
175 stop = translation.find('*', junc) |
4221664a2bd0
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/translate_bed_sequences commit 2a470e2c775a7427aa530e058510e4dc7b6d8e80"
galaxyp
parents:
0
diff
changeset
|
176 tstop = stop if stop >= 0 else len(translation) |
4221664a2bd0
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/translate_bed_sequences commit 2a470e2c775a7427aa530e058510e4dc7b6d8e80"
galaxyp
parents:
0
diff
changeset
|
177 offset = (block_sum - i) % 3 |
4221664a2bd0
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/translate_bed_sequences commit 2a470e2c775a7427aa530e058510e4dc7b6d8e80"
galaxyp
parents:
0
diff
changeset
|
178 trimmed = translation[tstart:tstop] |
4221664a2bd0
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/translate_bed_sequences commit 2a470e2c775a7427aa530e058510e4dc7b6d8e80"
galaxyp
parents:
0
diff
changeset
|
179 if debug: |
4221664a2bd0
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/translate_bed_sequences commit 2a470e2c775a7427aa530e058510e4dc7b6d8e80"
galaxyp
parents:
0
diff
changeset
|
180 sys.stderr.write("frame: %d\ttstart: %d tstop: %d offset: %d\t%s\n" % (i, tstart, tstop, offset, trimmed)) |
4221664a2bd0
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/translate_bed_sequences commit 2a470e2c775a7427aa530e058510e4dc7b6d8e80"
galaxyp
parents:
0
diff
changeset
|
181 if filtering and tstart > ignore: |
4221664a2bd0
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/translate_bed_sequences commit 2a470e2c775a7427aa530e058510e4dc7b6d8e80"
galaxyp
parents:
0
diff
changeset
|
182 continue |
4221664a2bd0
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/translate_bed_sequences commit 2a470e2c775a7427aa530e058510e4dc7b6d8e80"
galaxyp
parents:
0
diff
changeset
|
183 # get genomic locations for start and end |
4221664a2bd0
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/translate_bed_sequences commit 2a470e2c775a7427aa530e058510e4dc7b6d8e80"
galaxyp
parents:
0
diff
changeset
|
184 if self.strand == '+': |
4221664a2bd0
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/translate_bed_sequences commit 2a470e2c775a7427aa530e058510e4dc7b6d8e80"
galaxyp
parents:
0
diff
changeset
|
185 chromStart = self.chromStart + i + (tstart * 3) |
4221664a2bd0
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/translate_bed_sequences commit 2a470e2c775a7427aa530e058510e4dc7b6d8e80"
galaxyp
parents:
0
diff
changeset
|
186 chromEnd = self.chromEnd - offset - (len(translation) - tstop) * 3 |
4221664a2bd0
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/translate_bed_sequences commit 2a470e2c775a7427aa530e058510e4dc7b6d8e80"
galaxyp
parents:
0
diff
changeset
|
187 else: |
4221664a2bd0
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/translate_bed_sequences commit 2a470e2c775a7427aa530e058510e4dc7b6d8e80"
galaxyp
parents:
0
diff
changeset
|
188 chromStart = self.chromStart + offset + (len(translation) - tstop) * 3 |
4221664a2bd0
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/translate_bed_sequences commit 2a470e2c775a7427aa530e058510e4dc7b6d8e80"
galaxyp
parents:
0
diff
changeset
|
189 chromEnd = self.chromEnd - i - (tstart * 3) |
4221664a2bd0
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/translate_bed_sequences commit 2a470e2c775a7427aa530e058510e4dc7b6d8e80"
galaxyp
parents:
0
diff
changeset
|
190 # get the blocks for this translation |
4221664a2bd0
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/translate_bed_sequences commit 2a470e2c775a7427aa530e058510e4dc7b6d8e80"
galaxyp
parents:
0
diff
changeset
|
191 (tblockCount, tblockSizes, tblockStarts) = self.get_blocks(chromStart, chromEnd) |
4221664a2bd0
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/translate_bed_sequences commit 2a470e2c775a7427aa530e058510e4dc7b6d8e80"
galaxyp
parents:
0
diff
changeset
|
192 translations[i] = (chromStart, chromEnd, trimmed, tblockCount, tblockSizes, tblockStarts) |
4221664a2bd0
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/translate_bed_sequences commit 2a470e2c775a7427aa530e058510e4dc7b6d8e80"
galaxyp
parents:
0
diff
changeset
|
193 if debug: |
4221664a2bd0
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/translate_bed_sequences commit 2a470e2c775a7427aa530e058510e4dc7b6d8e80"
galaxyp
parents:
0
diff
changeset
|
194 sys.stderr.write("tblockCount: %d tblockStarts: %s tblockSizes: %s\n" % (tblockCount, tblockStarts, tblockSizes)) |
4221664a2bd0
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/translate_bed_sequences commit 2a470e2c775a7427aa530e058510e4dc7b6d8e80"
galaxyp
parents:
0
diff
changeset
|
195 # translations[i] = (chromStart, chromEnd, trimmed, tblockCount, tblockSizes, tblockStarts) |
4221664a2bd0
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/translate_bed_sequences commit 2a470e2c775a7427aa530e058510e4dc7b6d8e80"
galaxyp
parents:
0
diff
changeset
|
196 return translations |
4221664a2bd0
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/translate_bed_sequences commit 2a470e2c775a7427aa530e058510e4dc7b6d8e80"
galaxyp
parents:
0
diff
changeset
|
197 |
4221664a2bd0
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/translate_bed_sequences commit 2a470e2c775a7427aa530e058510e4dc7b6d8e80"
galaxyp
parents:
0
diff
changeset
|
198 def get_seq_id(self, seqtype='unk:unk', reference='', frame=None): |
4221664a2bd0
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/translate_bed_sequences commit 2a470e2c775a7427aa530e058510e4dc7b6d8e80"
galaxyp
parents:
0
diff
changeset
|
199 """ |
4221664a2bd0
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/translate_bed_sequences commit 2a470e2c775a7427aa530e058510e4dc7b6d8e80"
galaxyp
parents:
0
diff
changeset
|
200 # Ensembl fasta ID format |
4221664a2bd0
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/translate_bed_sequences commit 2a470e2c775a7427aa530e058510e4dc7b6d8e80"
galaxyp
parents:
0
diff
changeset
|
201 >ID SEQTYPE:STATUS LOCATION GENE TRANSCRIPT |
4221664a2bd0
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/translate_bed_sequences commit 2a470e2c775a7427aa530e058510e4dc7b6d8e80"
galaxyp
parents:
0
diff
changeset
|
202 >ENSP00000328693 pep:splice chromosome:NCBI35:1:904515:910768:1 gene:ENSG00000158815:transcript:ENST00000328693 gene_biotype:protein_coding transcript_biotype:protein_coding |
4221664a2bd0
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/translate_bed_sequences commit 2a470e2c775a7427aa530e058510e4dc7b6d8e80"
galaxyp
parents:
0
diff
changeset
|
203 """ |
4221664a2bd0
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/translate_bed_sequences commit 2a470e2c775a7427aa530e058510e4dc7b6d8e80"
galaxyp
parents:
0
diff
changeset
|
204 frame_name = '' |
4221664a2bd0
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/translate_bed_sequences commit 2a470e2c775a7427aa530e058510e4dc7b6d8e80"
galaxyp
parents:
0
diff
changeset
|
205 chromStart = self.chromStart |
4221664a2bd0
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/translate_bed_sequences commit 2a470e2c775a7427aa530e058510e4dc7b6d8e80"
galaxyp
parents:
0
diff
changeset
|
206 chromEnd = self.chromEnd |
4221664a2bd0
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/translate_bed_sequences commit 2a470e2c775a7427aa530e058510e4dc7b6d8e80"
galaxyp
parents:
0
diff
changeset
|
207 strand = 1 if self.strand == '+' else -1 |
4221664a2bd0
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/translate_bed_sequences commit 2a470e2c775a7427aa530e058510e4dc7b6d8e80"
galaxyp
parents:
0
diff
changeset
|
208 if frame is not None: |
4221664a2bd0
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/translate_bed_sequences commit 2a470e2c775a7427aa530e058510e4dc7b6d8e80"
galaxyp
parents:
0
diff
changeset
|
209 block_sum = sum(self.blockSizes) |
4221664a2bd0
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/translate_bed_sequences commit 2a470e2c775a7427aa530e058510e4dc7b6d8e80"
galaxyp
parents:
0
diff
changeset
|
210 offset = (block_sum - frame) % 3 |
4221664a2bd0
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/translate_bed_sequences commit 2a470e2c775a7427aa530e058510e4dc7b6d8e80"
galaxyp
parents:
0
diff
changeset
|
211 frame_name = '_' + str(frame + 1) |
4221664a2bd0
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/translate_bed_sequences commit 2a470e2c775a7427aa530e058510e4dc7b6d8e80"
galaxyp
parents:
0
diff
changeset
|
212 if self.strand == '+': |
4221664a2bd0
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/translate_bed_sequences commit 2a470e2c775a7427aa530e058510e4dc7b6d8e80"
galaxyp
parents:
0
diff
changeset
|
213 chromStart += frame |
4221664a2bd0
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/translate_bed_sequences commit 2a470e2c775a7427aa530e058510e4dc7b6d8e80"
galaxyp
parents:
0
diff
changeset
|
214 chromEnd -= offset |
4221664a2bd0
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/translate_bed_sequences commit 2a470e2c775a7427aa530e058510e4dc7b6d8e80"
galaxyp
parents:
0
diff
changeset
|
215 else: |
4221664a2bd0
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/translate_bed_sequences commit 2a470e2c775a7427aa530e058510e4dc7b6d8e80"
galaxyp
parents:
0
diff
changeset
|
216 chromStart += offset |
4221664a2bd0
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/translate_bed_sequences commit 2a470e2c775a7427aa530e058510e4dc7b6d8e80"
galaxyp
parents:
0
diff
changeset
|
217 chromEnd -= frame |
4221664a2bd0
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/translate_bed_sequences commit 2a470e2c775a7427aa530e058510e4dc7b6d8e80"
galaxyp
parents:
0
diff
changeset
|
218 location = "chromosome:%s:%s:%s:%s:%s" % (reference, self.chrom, chromStart, chromEnd, strand) |
4221664a2bd0
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/translate_bed_sequences commit 2a470e2c775a7427aa530e058510e4dc7b6d8e80"
galaxyp
parents:
0
diff
changeset
|
219 seq_id = "%s%s %s %s" % (self.name, frame_name, seqtype, location) |
4221664a2bd0
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/translate_bed_sequences commit 2a470e2c775a7427aa530e058510e4dc7b6d8e80"
galaxyp
parents:
0
diff
changeset
|
220 return seq_id |
4221664a2bd0
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/translate_bed_sequences commit 2a470e2c775a7427aa530e058510e4dc7b6d8e80"
galaxyp
parents:
0
diff
changeset
|
221 |
4221664a2bd0
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/translate_bed_sequences commit 2a470e2c775a7427aa530e058510e4dc7b6d8e80"
galaxyp
parents:
0
diff
changeset
|
222 def get_line(self, start_offset=0, end_offset=0): |
4221664a2bd0
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/translate_bed_sequences commit 2a470e2c775a7427aa530e058510e4dc7b6d8e80"
galaxyp
parents:
0
diff
changeset
|
223 if start_offset or end_offset: |
4221664a2bd0
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/translate_bed_sequences commit 2a470e2c775a7427aa530e058510e4dc7b6d8e80"
galaxyp
parents:
0
diff
changeset
|
224 s_offset = start_offset if start_offset else 0 |
4221664a2bd0
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/translate_bed_sequences commit 2a470e2c775a7427aa530e058510e4dc7b6d8e80"
galaxyp
parents:
0
diff
changeset
|
225 e_offset = end_offset if end_offset else 0 |
4221664a2bd0
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/translate_bed_sequences commit 2a470e2c775a7427aa530e058510e4dc7b6d8e80"
galaxyp
parents:
0
diff
changeset
|
226 if s_offset > self.chromStart: |
4221664a2bd0
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/translate_bed_sequences commit 2a470e2c775a7427aa530e058510e4dc7b6d8e80"
galaxyp
parents:
0
diff
changeset
|
227 s_offset = self.chromStart |
4221664a2bd0
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/translate_bed_sequences commit 2a470e2c775a7427aa530e058510e4dc7b6d8e80"
galaxyp
parents:
0
diff
changeset
|
228 chrStart = self.chromStart - s_offset |
4221664a2bd0
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/translate_bed_sequences commit 2a470e2c775a7427aa530e058510e4dc7b6d8e80"
galaxyp
parents:
0
diff
changeset
|
229 chrEnd = self.chromEnd + e_offset |
4221664a2bd0
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/translate_bed_sequences commit 2a470e2c775a7427aa530e058510e4dc7b6d8e80"
galaxyp
parents:
0
diff
changeset
|
230 blkSizes = self.blockSizes |
4221664a2bd0
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/translate_bed_sequences commit 2a470e2c775a7427aa530e058510e4dc7b6d8e80"
galaxyp
parents:
0
diff
changeset
|
231 blkSizes[0] += s_offset |
4221664a2bd0
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/translate_bed_sequences commit 2a470e2c775a7427aa530e058510e4dc7b6d8e80"
galaxyp
parents:
0
diff
changeset
|
232 blkSizes[-1] += e_offset |
4221664a2bd0
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/translate_bed_sequences commit 2a470e2c775a7427aa530e058510e4dc7b6d8e80"
galaxyp
parents:
0
diff
changeset
|
233 blkStarts = self.blockStarts |
4221664a2bd0
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/translate_bed_sequences commit 2a470e2c775a7427aa530e058510e4dc7b6d8e80"
galaxyp
parents:
0
diff
changeset
|
234 for i in range(1, self.blockCount): |
4221664a2bd0
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/translate_bed_sequences commit 2a470e2c775a7427aa530e058510e4dc7b6d8e80"
galaxyp
parents:
0
diff
changeset
|
235 blkStarts[i] += s_offset |
4221664a2bd0
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/translate_bed_sequences commit 2a470e2c775a7427aa530e058510e4dc7b6d8e80"
galaxyp
parents:
0
diff
changeset
|
236 items = [str(x) for x in [self.chrom, chrStart, chrEnd, self.name, self.score, self.strand, self.thickStart, self.thickEnd, self.itemRgb, self.blockCount, ','.join([str(x) for x in blkSizes]), ','.join([str(x) for x in blkStarts])]] |
4221664a2bd0
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/translate_bed_sequences commit 2a470e2c775a7427aa530e058510e4dc7b6d8e80"
galaxyp
parents:
0
diff
changeset
|
237 return '\t'.join(items) + '\n' |
4221664a2bd0
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/translate_bed_sequences commit 2a470e2c775a7427aa530e058510e4dc7b6d8e80"
galaxyp
parents:
0
diff
changeset
|
238 return self.line |
4221664a2bd0
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/translate_bed_sequences commit 2a470e2c775a7427aa530e058510e4dc7b6d8e80"
galaxyp
parents:
0
diff
changeset
|
239 |
0
d723eb657f1d
planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/translate_bed_sequences commit e04ed4b4960d6109a85c1cc68a2bf4931c8751ef-dirty
galaxyp
parents:
diff
changeset
|
240 |
d723eb657f1d
planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/translate_bed_sequences commit e04ed4b4960d6109a85c1cc68a2bf4931c8751ef-dirty
galaxyp
parents:
diff
changeset
|
241 def __main__(): |
2
4221664a2bd0
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/translate_bed_sequences commit 2a470e2c775a7427aa530e058510e4dc7b6d8e80"
galaxyp
parents:
0
diff
changeset
|
242 # Parse Command Line |
4221664a2bd0
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/translate_bed_sequences commit 2a470e2c775a7427aa530e058510e4dc7b6d8e80"
galaxyp
parents:
0
diff
changeset
|
243 parser = optparse.OptionParser() |
4221664a2bd0
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/translate_bed_sequences commit 2a470e2c775a7427aa530e058510e4dc7b6d8e80"
galaxyp
parents:
0
diff
changeset
|
244 parser.add_option('-i', '--input', dest='input', help='BED file (tophat junctions.bed) with sequence column added') |
4221664a2bd0
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/translate_bed_sequences commit 2a470e2c775a7427aa530e058510e4dc7b6d8e80"
galaxyp
parents:
0
diff
changeset
|
245 parser.add_option('-o', '--output', dest='output', help='Translations of spliced sequence') |
4221664a2bd0
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/translate_bed_sequences commit 2a470e2c775a7427aa530e058510e4dc7b6d8e80"
galaxyp
parents:
0
diff
changeset
|
246 parser.add_option('-b', '--bed_format', dest='bed_format', action='store_true', default=False, help='Append translations to bed file instead of fasta') |
4221664a2bd0
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/translate_bed_sequences commit 2a470e2c775a7427aa530e058510e4dc7b6d8e80"
galaxyp
parents:
0
diff
changeset
|
247 parser.add_option('-D', '--fa_db', dest='fa_db', default=None, help='Prefix DB identifier for fasta ID line, e.g. generic') |
4221664a2bd0
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/translate_bed_sequences commit 2a470e2c775a7427aa530e058510e4dc7b6d8e80"
galaxyp
parents:
0
diff
changeset
|
248 parser.add_option('-s', '--fa_sep', dest='fa_sep', default='|', help='fasta ID separator defaults to pipe char, e.g. generic|ProtID|description') |
4221664a2bd0
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/translate_bed_sequences commit 2a470e2c775a7427aa530e058510e4dc7b6d8e80"
galaxyp
parents:
0
diff
changeset
|
249 parser.add_option('-B', '--bed', dest='bed', default=None, help='Output a bed file with added 13th column having translation') |
4221664a2bd0
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/translate_bed_sequences commit 2a470e2c775a7427aa530e058510e4dc7b6d8e80"
galaxyp
parents:
0
diff
changeset
|
250 parser.add_option('-G', '--gff3', dest='gff', default=None, help='Output translations to a GFF3 file') |
4221664a2bd0
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/translate_bed_sequences commit 2a470e2c775a7427aa530e058510e4dc7b6d8e80"
galaxyp
parents:
0
diff
changeset
|
251 parser.add_option('-S', '--seqtype', dest='seqtype', default='pep:splice', help='SEQTYPE:STATUS for fasta ID line') |
4221664a2bd0
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/translate_bed_sequences commit 2a470e2c775a7427aa530e058510e4dc7b6d8e80"
galaxyp
parents:
0
diff
changeset
|
252 parser.add_option('-P', '--id_prefix', dest='id_prefix', default='', help='prefix for the sequence ID') |
4221664a2bd0
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/translate_bed_sequences commit 2a470e2c775a7427aa530e058510e4dc7b6d8e80"
galaxyp
parents:
0
diff
changeset
|
253 parser.add_option('-R', '--reference', dest='reference', default=None, help='Genome Reference Name for fasta ID location ') |
4221664a2bd0
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/translate_bed_sequences commit 2a470e2c775a7427aa530e058510e4dc7b6d8e80"
galaxyp
parents:
0
diff
changeset
|
254 parser.add_option('-r', '--refsource', dest='refsource', default=None, help='Source for Genome Reference, e.g. Ensembl, UCSC, or NCBI') |
4221664a2bd0
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/translate_bed_sequences commit 2a470e2c775a7427aa530e058510e4dc7b6d8e80"
galaxyp
parents:
0
diff
changeset
|
255 parser.add_option('-Q', '--score_name', dest='score_name', default=None, help='include in the fasta ID line score_name:score ') |
4221664a2bd0
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/translate_bed_sequences commit 2a470e2c775a7427aa530e058510e4dc7b6d8e80"
galaxyp
parents:
0
diff
changeset
|
256 parser.add_option('-l', '--leading_bp', dest='leading_bp', type='int', default=None, help='leading number of base pairs to ignore when filtering') |
4221664a2bd0
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/translate_bed_sequences commit 2a470e2c775a7427aa530e058510e4dc7b6d8e80"
galaxyp
parents:
0
diff
changeset
|
257 parser.add_option('-t', '--trailing_bp', dest='trailing_bp', type='int', default=None, help='trailing number of base pairs to ignore when filtering') |
4221664a2bd0
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/translate_bed_sequences commit 2a470e2c775a7427aa530e058510e4dc7b6d8e80"
galaxyp
parents:
0
diff
changeset
|
258 parser.add_option('-U', '--unfiltered', dest='filtering', action='store_false', default=True, help='Do NOT filterout translation with stop codon in the first exon') |
4221664a2bd0
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/translate_bed_sequences commit 2a470e2c775a7427aa530e058510e4dc7b6d8e80"
galaxyp
parents:
0
diff
changeset
|
259 parser.add_option('-u', '--untrimmed', dest='untrimmed', action='store_true', default=False, help='Do NOT trim from splice site to stop codon') |
4221664a2bd0
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/translate_bed_sequences commit 2a470e2c775a7427aa530e058510e4dc7b6d8e80"
galaxyp
parents:
0
diff
changeset
|
260 parser.add_option('-L', '--min_length', dest='min_length', type='int', default=None, help='Minimun length (to first stop codon)') |
4221664a2bd0
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/translate_bed_sequences commit 2a470e2c775a7427aa530e058510e4dc7b6d8e80"
galaxyp
parents:
0
diff
changeset
|
261 parser.add_option('-M', '--max_stop_codons', dest='max_stop_codons', type='int', default=None, help='Filter out translations with more than max_stop_codons') |
4221664a2bd0
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/translate_bed_sequences commit 2a470e2c775a7427aa530e058510e4dc7b6d8e80"
galaxyp
parents:
0
diff
changeset
|
262 parser.add_option('-d', '--debug', dest='debug', action='store_true', default=False, help='Turn on wrapper debugging to stdout') |
4221664a2bd0
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/translate_bed_sequences commit 2a470e2c775a7427aa530e058510e4dc7b6d8e80"
galaxyp
parents:
0
diff
changeset
|
263 (options, args) = parser.parse_args() |
4221664a2bd0
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/translate_bed_sequences commit 2a470e2c775a7427aa530e058510e4dc7b6d8e80"
galaxyp
parents:
0
diff
changeset
|
264 # Input files |
4221664a2bd0
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/translate_bed_sequences commit 2a470e2c775a7427aa530e058510e4dc7b6d8e80"
galaxyp
parents:
0
diff
changeset
|
265 if options.input is not None: |
4221664a2bd0
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/translate_bed_sequences commit 2a470e2c775a7427aa530e058510e4dc7b6d8e80"
galaxyp
parents:
0
diff
changeset
|
266 try: |
4221664a2bd0
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/translate_bed_sequences commit 2a470e2c775a7427aa530e058510e4dc7b6d8e80"
galaxyp
parents:
0
diff
changeset
|
267 inputPath = os.path.abspath(options.input) |
4221664a2bd0
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/translate_bed_sequences commit 2a470e2c775a7427aa530e058510e4dc7b6d8e80"
galaxyp
parents:
0
diff
changeset
|
268 inputFile = open(inputPath, 'r') |
4221664a2bd0
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/translate_bed_sequences commit 2a470e2c775a7427aa530e058510e4dc7b6d8e80"
galaxyp
parents:
0
diff
changeset
|
269 except Exception as e: |
4221664a2bd0
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/translate_bed_sequences commit 2a470e2c775a7427aa530e058510e4dc7b6d8e80"
galaxyp
parents:
0
diff
changeset
|
270 sys.stderr.write("failed: %s\n" % e) |
4221664a2bd0
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/translate_bed_sequences commit 2a470e2c775a7427aa530e058510e4dc7b6d8e80"
galaxyp
parents:
0
diff
changeset
|
271 exit(2) |
0
d723eb657f1d
planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/translate_bed_sequences commit e04ed4b4960d6109a85c1cc68a2bf4931c8751ef-dirty
galaxyp
parents:
diff
changeset
|
272 else: |
2
4221664a2bd0
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/translate_bed_sequences commit 2a470e2c775a7427aa530e058510e4dc7b6d8e80"
galaxyp
parents:
0
diff
changeset
|
273 inputFile = sys.stdin |
4221664a2bd0
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/translate_bed_sequences commit 2a470e2c775a7427aa530e058510e4dc7b6d8e80"
galaxyp
parents:
0
diff
changeset
|
274 # Output files |
4221664a2bd0
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/translate_bed_sequences commit 2a470e2c775a7427aa530e058510e4dc7b6d8e80"
galaxyp
parents:
0
diff
changeset
|
275 bed_fh = None |
4221664a2bd0
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/translate_bed_sequences commit 2a470e2c775a7427aa530e058510e4dc7b6d8e80"
galaxyp
parents:
0
diff
changeset
|
276 gff_fh = None |
4221664a2bd0
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/translate_bed_sequences commit 2a470e2c775a7427aa530e058510e4dc7b6d8e80"
galaxyp
parents:
0
diff
changeset
|
277 gff_fa_file = None |
4221664a2bd0
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/translate_bed_sequences commit 2a470e2c775a7427aa530e058510e4dc7b6d8e80"
galaxyp
parents:
0
diff
changeset
|
278 gff_fa = None |
4221664a2bd0
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/translate_bed_sequences commit 2a470e2c775a7427aa530e058510e4dc7b6d8e80"
galaxyp
parents:
0
diff
changeset
|
279 outFile = None |
4221664a2bd0
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/translate_bed_sequences commit 2a470e2c775a7427aa530e058510e4dc7b6d8e80"
galaxyp
parents:
0
diff
changeset
|
280 if options.output is None: |
4221664a2bd0
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/translate_bed_sequences commit 2a470e2c775a7427aa530e058510e4dc7b6d8e80"
galaxyp
parents:
0
diff
changeset
|
281 # write to stdout |
4221664a2bd0
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/translate_bed_sequences commit 2a470e2c775a7427aa530e058510e4dc7b6d8e80"
galaxyp
parents:
0
diff
changeset
|
282 outFile = sys.stdout |
4221664a2bd0
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/translate_bed_sequences commit 2a470e2c775a7427aa530e058510e4dc7b6d8e80"
galaxyp
parents:
0
diff
changeset
|
283 if options.gff: |
4221664a2bd0
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/translate_bed_sequences commit 2a470e2c775a7427aa530e058510e4dc7b6d8e80"
galaxyp
parents:
0
diff
changeset
|
284 gff_fa_file = tempfile.NamedTemporaryFile(prefix='gff_fasta_', suffix=".fa", dir=os.getcwd()).name |
4221664a2bd0
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/translate_bed_sequences commit 2a470e2c775a7427aa530e058510e4dc7b6d8e80"
galaxyp
parents:
0
diff
changeset
|
285 gff_fa = open(gff_fa_file, 'w') |
0
d723eb657f1d
planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/translate_bed_sequences commit e04ed4b4960d6109a85c1cc68a2bf4931c8751ef-dirty
galaxyp
parents:
diff
changeset
|
286 else: |
2
4221664a2bd0
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/translate_bed_sequences commit 2a470e2c775a7427aa530e058510e4dc7b6d8e80"
galaxyp
parents:
0
diff
changeset
|
287 try: |
4221664a2bd0
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/translate_bed_sequences commit 2a470e2c775a7427aa530e058510e4dc7b6d8e80"
galaxyp
parents:
0
diff
changeset
|
288 outPath = os.path.abspath(options.output) |
4221664a2bd0
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/translate_bed_sequences commit 2a470e2c775a7427aa530e058510e4dc7b6d8e80"
galaxyp
parents:
0
diff
changeset
|
289 outFile = open(outPath, 'w') |
4221664a2bd0
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/translate_bed_sequences commit 2a470e2c775a7427aa530e058510e4dc7b6d8e80"
galaxyp
parents:
0
diff
changeset
|
290 except Exception as e: |
4221664a2bd0
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/translate_bed_sequences commit 2a470e2c775a7427aa530e058510e4dc7b6d8e80"
galaxyp
parents:
0
diff
changeset
|
291 sys.stderr.write("failed: %s\n" % e) |
4221664a2bd0
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/translate_bed_sequences commit 2a470e2c775a7427aa530e058510e4dc7b6d8e80"
galaxyp
parents:
0
diff
changeset
|
292 exit(3) |
4221664a2bd0
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/translate_bed_sequences commit 2a470e2c775a7427aa530e058510e4dc7b6d8e80"
galaxyp
parents:
0
diff
changeset
|
293 if options.gff: |
4221664a2bd0
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/translate_bed_sequences commit 2a470e2c775a7427aa530e058510e4dc7b6d8e80"
galaxyp
parents:
0
diff
changeset
|
294 gff_fa_file = outPath |
4221664a2bd0
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/translate_bed_sequences commit 2a470e2c775a7427aa530e058510e4dc7b6d8e80"
galaxyp
parents:
0
diff
changeset
|
295 if options.bed: |
4221664a2bd0
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/translate_bed_sequences commit 2a470e2c775a7427aa530e058510e4dc7b6d8e80"
galaxyp
parents:
0
diff
changeset
|
296 bed_fh = open(options.bed, 'w') |
4221664a2bd0
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/translate_bed_sequences commit 2a470e2c775a7427aa530e058510e4dc7b6d8e80"
galaxyp
parents:
0
diff
changeset
|
297 bed_fh.write('track name="%s" description="%s" \n' % ('novel_junctioni_translations', 'test')) |
4221664a2bd0
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/translate_bed_sequences commit 2a470e2c775a7427aa530e058510e4dc7b6d8e80"
galaxyp
parents:
0
diff
changeset
|
298 if options.gff: |
4221664a2bd0
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/translate_bed_sequences commit 2a470e2c775a7427aa530e058510e4dc7b6d8e80"
galaxyp
parents:
0
diff
changeset
|
299 gff_fh = open(options.gff, 'w') |
4221664a2bd0
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/translate_bed_sequences commit 2a470e2c775a7427aa530e058510e4dc7b6d8e80"
galaxyp
parents:
0
diff
changeset
|
300 gff_fh.write("##gff-version 3.2.1\n") |
4221664a2bd0
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/translate_bed_sequences commit 2a470e2c775a7427aa530e058510e4dc7b6d8e80"
galaxyp
parents:
0
diff
changeset
|
301 if options.reference: |
4221664a2bd0
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/translate_bed_sequences commit 2a470e2c775a7427aa530e058510e4dc7b6d8e80"
galaxyp
parents:
0
diff
changeset
|
302 gff_fh.write("##genome-build %s %s\n" % (options.refsource if options.refsource else 'unknown', options.reference)) |
4221664a2bd0
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/translate_bed_sequences commit 2a470e2c775a7427aa530e058510e4dc7b6d8e80"
galaxyp
parents:
0
diff
changeset
|
303 leading_bp = 0 |
4221664a2bd0
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/translate_bed_sequences commit 2a470e2c775a7427aa530e058510e4dc7b6d8e80"
galaxyp
parents:
0
diff
changeset
|
304 trailing_bp = 0 |
4221664a2bd0
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/translate_bed_sequences commit 2a470e2c775a7427aa530e058510e4dc7b6d8e80"
galaxyp
parents:
0
diff
changeset
|
305 if options.leading_bp: |
4221664a2bd0
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/translate_bed_sequences commit 2a470e2c775a7427aa530e058510e4dc7b6d8e80"
galaxyp
parents:
0
diff
changeset
|
306 if options.leading_bp >= 0: |
4221664a2bd0
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/translate_bed_sequences commit 2a470e2c775a7427aa530e058510e4dc7b6d8e80"
galaxyp
parents:
0
diff
changeset
|
307 leading_bp = options.leading_bp |
4221664a2bd0
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/translate_bed_sequences commit 2a470e2c775a7427aa530e058510e4dc7b6d8e80"
galaxyp
parents:
0
diff
changeset
|
308 else: |
4221664a2bd0
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/translate_bed_sequences commit 2a470e2c775a7427aa530e058510e4dc7b6d8e80"
galaxyp
parents:
0
diff
changeset
|
309 sys.stderr.write("failed: leading_bp must be positive\n") |
4221664a2bd0
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/translate_bed_sequences commit 2a470e2c775a7427aa530e058510e4dc7b6d8e80"
galaxyp
parents:
0
diff
changeset
|
310 exit(5) |
4221664a2bd0
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/translate_bed_sequences commit 2a470e2c775a7427aa530e058510e4dc7b6d8e80"
galaxyp
parents:
0
diff
changeset
|
311 if options.trailing_bp: |
4221664a2bd0
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/translate_bed_sequences commit 2a470e2c775a7427aa530e058510e4dc7b6d8e80"
galaxyp
parents:
0
diff
changeset
|
312 if options.trailing_bp >= 0: |
4221664a2bd0
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/translate_bed_sequences commit 2a470e2c775a7427aa530e058510e4dc7b6d8e80"
galaxyp
parents:
0
diff
changeset
|
313 trailing_bp = options.trailing_bp |
4221664a2bd0
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/translate_bed_sequences commit 2a470e2c775a7427aa530e058510e4dc7b6d8e80"
galaxyp
parents:
0
diff
changeset
|
314 else: |
4221664a2bd0
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/translate_bed_sequences commit 2a470e2c775a7427aa530e058510e4dc7b6d8e80"
galaxyp
parents:
0
diff
changeset
|
315 sys.stderr.write("failed: trailing_bp must be positive\n") |
4221664a2bd0
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/translate_bed_sequences commit 2a470e2c775a7427aa530e058510e4dc7b6d8e80"
galaxyp
parents:
0
diff
changeset
|
316 exit(5) |
4221664a2bd0
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/translate_bed_sequences commit 2a470e2c775a7427aa530e058510e4dc7b6d8e80"
galaxyp
parents:
0
diff
changeset
|
317 # Scan bed file |
4221664a2bd0
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/translate_bed_sequences commit 2a470e2c775a7427aa530e058510e4dc7b6d8e80"
galaxyp
parents:
0
diff
changeset
|
318 try: |
4221664a2bd0
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/translate_bed_sequences commit 2a470e2c775a7427aa530e058510e4dc7b6d8e80"
galaxyp
parents:
0
diff
changeset
|
319 for i, line in enumerate(inputFile): |
4221664a2bd0
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/translate_bed_sequences commit 2a470e2c775a7427aa530e058510e4dc7b6d8e80"
galaxyp
parents:
0
diff
changeset
|
320 if line.startswith('track'): |
4221664a2bd0
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/translate_bed_sequences commit 2a470e2c775a7427aa530e058510e4dc7b6d8e80"
galaxyp
parents:
0
diff
changeset
|
321 if outFile and options.bed_format: |
4221664a2bd0
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/translate_bed_sequences commit 2a470e2c775a7427aa530e058510e4dc7b6d8e80"
galaxyp
parents:
0
diff
changeset
|
322 outFile.write(line) |
4221664a2bd0
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/translate_bed_sequences commit 2a470e2c775a7427aa530e058510e4dc7b6d8e80"
galaxyp
parents:
0
diff
changeset
|
323 continue |
4221664a2bd0
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/translate_bed_sequences commit 2a470e2c775a7427aa530e058510e4dc7b6d8e80"
galaxyp
parents:
0
diff
changeset
|
324 entry = BedEntry(line) |
4221664a2bd0
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/translate_bed_sequences commit 2a470e2c775a7427aa530e058510e4dc7b6d8e80"
galaxyp
parents:
0
diff
changeset
|
325 strand = 1 if entry.strand == '+' else -1 |
4221664a2bd0
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/translate_bed_sequences commit 2a470e2c775a7427aa530e058510e4dc7b6d8e80"
galaxyp
parents:
0
diff
changeset
|
326 translations = entry.get_translations() |
4221664a2bd0
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/translate_bed_sequences commit 2a470e2c775a7427aa530e058510e4dc7b6d8e80"
galaxyp
parents:
0
diff
changeset
|
327 if options.debug: |
4221664a2bd0
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/translate_bed_sequences commit 2a470e2c775a7427aa530e058510e4dc7b6d8e80"
galaxyp
parents:
0
diff
changeset
|
328 exon_seqs = entry.get_exon_seqs() |
4221664a2bd0
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/translate_bed_sequences commit 2a470e2c775a7427aa530e058510e4dc7b6d8e80"
galaxyp
parents:
0
diff
changeset
|
329 exon_sizes = [len(seq) for seq in exon_seqs] |
4221664a2bd0
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/translate_bed_sequences commit 2a470e2c775a7427aa530e058510e4dc7b6d8e80"
galaxyp
parents:
0
diff
changeset
|
330 splice_sites = [int(sum(exon_sizes[:x]) / 3) for x in range(1, len(exon_sizes))] |
4221664a2bd0
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/translate_bed_sequences commit 2a470e2c775a7427aa530e058510e4dc7b6d8e80"
galaxyp
parents:
0
diff
changeset
|
331 sys.stderr.write("%s\n" % entry.name) |
4221664a2bd0
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/translate_bed_sequences commit 2a470e2c775a7427aa530e058510e4dc7b6d8e80"
galaxyp
parents:
0
diff
changeset
|
332 sys.stderr.write("%s\n" % line.rstrip('\r\n')) |
4221664a2bd0
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/translate_bed_sequences commit 2a470e2c775a7427aa530e058510e4dc7b6d8e80"
galaxyp
parents:
0
diff
changeset
|
333 sys.stderr.write("exons: %s\n" % exon_seqs) |
4221664a2bd0
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/translate_bed_sequences commit 2a470e2c775a7427aa530e058510e4dc7b6d8e80"
galaxyp
parents:
0
diff
changeset
|
334 sys.stderr.write("%s\n" % splice_sites) |
4221664a2bd0
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/translate_bed_sequences commit 2a470e2c775a7427aa530e058510e4dc7b6d8e80"
galaxyp
parents:
0
diff
changeset
|
335 for i, translation in enumerate(translations): |
4221664a2bd0
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/translate_bed_sequences commit 2a470e2c775a7427aa530e058510e4dc7b6d8e80"
galaxyp
parents:
0
diff
changeset
|
336 sys.stderr.write("frame %d: %s\n" % (i + 1, translation)) |
4221664a2bd0
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/translate_bed_sequences commit 2a470e2c775a7427aa530e058510e4dc7b6d8e80"
galaxyp
parents:
0
diff
changeset
|
337 sys.stderr.write("splice: %s\n" % (''.join(['^' if int(((j * 3) + i) / 3) in splice_sites else '-' for j in range(len(translation))]))) |
4221664a2bd0
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/translate_bed_sequences commit 2a470e2c775a7427aa530e058510e4dc7b6d8e80"
galaxyp
parents:
0
diff
changeset
|
338 sys.stderr.write("\n") |
4221664a2bd0
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/translate_bed_sequences commit 2a470e2c775a7427aa530e058510e4dc7b6d8e80"
galaxyp
parents:
0
diff
changeset
|
339 if options.bed_format: |
4221664a2bd0
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/translate_bed_sequences commit 2a470e2c775a7427aa530e058510e4dc7b6d8e80"
galaxyp
parents:
0
diff
changeset
|
340 tx_entry = "%s\t%s\n" % (line.rstrip('\r\n'), '\t'.join(translations)) |
4221664a2bd0
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/translate_bed_sequences commit 2a470e2c775a7427aa530e058510e4dc7b6d8e80"
galaxyp
parents:
0
diff
changeset
|
341 outFile.write(tx_entry) |
4221664a2bd0
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/translate_bed_sequences commit 2a470e2c775a7427aa530e058510e4dc7b6d8e80"
galaxyp
parents:
0
diff
changeset
|
342 else: |
4221664a2bd0
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/translate_bed_sequences commit 2a470e2c775a7427aa530e058510e4dc7b6d8e80"
galaxyp
parents:
0
diff
changeset
|
343 translations = entry.get_filterd_translations(untrimmed=options.untrimmed, filtering=options.filtering, ignore_left_bp=leading_bp, ignore_right_bp=trailing_bp, debug=options.debug) |
4221664a2bd0
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/translate_bed_sequences commit 2a470e2c775a7427aa530e058510e4dc7b6d8e80"
galaxyp
parents:
0
diff
changeset
|
344 for i, tx in enumerate(translations): |
4221664a2bd0
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/translate_bed_sequences commit 2a470e2c775a7427aa530e058510e4dc7b6d8e80"
galaxyp
parents:
0
diff
changeset
|
345 if tx: |
4221664a2bd0
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/translate_bed_sequences commit 2a470e2c775a7427aa530e058510e4dc7b6d8e80"
galaxyp
parents:
0
diff
changeset
|
346 (chromStart, chromEnd, translation, blockCount, blockSizes, blockStarts) = tx |
4221664a2bd0
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/translate_bed_sequences commit 2a470e2c775a7427aa530e058510e4dc7b6d8e80"
galaxyp
parents:
0
diff
changeset
|
347 if options.min_length is not None and len(translation) < options.min_length: |
4221664a2bd0
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/translate_bed_sequences commit 2a470e2c775a7427aa530e058510e4dc7b6d8e80"
galaxyp
parents:
0
diff
changeset
|
348 continue |
4221664a2bd0
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/translate_bed_sequences commit 2a470e2c775a7427aa530e058510e4dc7b6d8e80"
galaxyp
parents:
0
diff
changeset
|
349 if options.max_stop_codons is not None and translation.count('*') > options.max_stop_codons: |
4221664a2bd0
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/translate_bed_sequences commit 2a470e2c775a7427aa530e058510e4dc7b6d8e80"
galaxyp
parents:
0
diff
changeset
|
350 continue |
4221664a2bd0
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/translate_bed_sequences commit 2a470e2c775a7427aa530e058510e4dc7b6d8e80"
galaxyp
parents:
0
diff
changeset
|
351 frame_name = '_%s' % (i + 1) |
4221664a2bd0
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/translate_bed_sequences commit 2a470e2c775a7427aa530e058510e4dc7b6d8e80"
galaxyp
parents:
0
diff
changeset
|
352 pep_id = "%s%s%s" % (options.id_prefix, entry.name, frame_name) |
4221664a2bd0
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/translate_bed_sequences commit 2a470e2c775a7427aa530e058510e4dc7b6d8e80"
galaxyp
parents:
0
diff
changeset
|
353 if bed_fh: |
4221664a2bd0
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/translate_bed_sequences commit 2a470e2c775a7427aa530e058510e4dc7b6d8e80"
galaxyp
parents:
0
diff
changeset
|
354 bed_fh.write('%s\t%d\t%d\t%s\t%d\t%s\t%d\t%d\t%s\t%d\t%s\t%s\t%s\n' % (str(entry.chrom), chromStart, chromEnd, pep_id, entry.score, entry.strand, chromStart, chromEnd, entry.itemRgb, blockCount, ','.join([str(x) for x in blockSizes]), ','.join([str(x) for x in blockStarts]), translation)) |
4221664a2bd0
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/translate_bed_sequences commit 2a470e2c775a7427aa530e058510e4dc7b6d8e80"
galaxyp
parents:
0
diff
changeset
|
355 location = "chromosome:%s:%s:%s:%s:%s" % (options.reference, entry.chrom, chromStart, chromEnd, strand) |
4221664a2bd0
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/translate_bed_sequences commit 2a470e2c775a7427aa530e058510e4dc7b6d8e80"
galaxyp
parents:
0
diff
changeset
|
356 score = " %s:%s" % (options.score_name, entry.score) if options.score_name else '' |
4221664a2bd0
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/translate_bed_sequences commit 2a470e2c775a7427aa530e058510e4dc7b6d8e80"
galaxyp
parents:
0
diff
changeset
|
357 seq_description = "%s %s%s" % (options.seqtype, location, score) |
4221664a2bd0
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/translate_bed_sequences commit 2a470e2c775a7427aa530e058510e4dc7b6d8e80"
galaxyp
parents:
0
diff
changeset
|
358 seq_id = "%s " % pep_id |
4221664a2bd0
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/translate_bed_sequences commit 2a470e2c775a7427aa530e058510e4dc7b6d8e80"
galaxyp
parents:
0
diff
changeset
|
359 if options.fa_db: |
4221664a2bd0
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/translate_bed_sequences commit 2a470e2c775a7427aa530e058510e4dc7b6d8e80"
galaxyp
parents:
0
diff
changeset
|
360 seq_id = "%s%s%s%s" % (options.fa_db, options.fa_sep, pep_id, options.fa_sep) |
4221664a2bd0
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/translate_bed_sequences commit 2a470e2c775a7427aa530e058510e4dc7b6d8e80"
galaxyp
parents:
0
diff
changeset
|
361 fa_id = "%s%s" % (seq_id, seq_description) |
4221664a2bd0
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/translate_bed_sequences commit 2a470e2c775a7427aa530e058510e4dc7b6d8e80"
galaxyp
parents:
0
diff
changeset
|
362 fa_entry = ">%s\n%s\n" % (fa_id, translation) |
4221664a2bd0
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/translate_bed_sequences commit 2a470e2c775a7427aa530e058510e4dc7b6d8e80"
galaxyp
parents:
0
diff
changeset
|
363 outFile.write(fa_entry) |
4221664a2bd0
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/translate_bed_sequences commit 2a470e2c775a7427aa530e058510e4dc7b6d8e80"
galaxyp
parents:
0
diff
changeset
|
364 if gff_fh: |
4221664a2bd0
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/translate_bed_sequences commit 2a470e2c775a7427aa530e058510e4dc7b6d8e80"
galaxyp
parents:
0
diff
changeset
|
365 if gff_fa: |
4221664a2bd0
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/translate_bed_sequences commit 2a470e2c775a7427aa530e058510e4dc7b6d8e80"
galaxyp
parents:
0
diff
changeset
|
366 gff_fa.write(fa_entry) |
4221664a2bd0
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/translate_bed_sequences commit 2a470e2c775a7427aa530e058510e4dc7b6d8e80"
galaxyp
parents:
0
diff
changeset
|
367 gff_fh.write("##sequence-region %s %d %d\n" % (entry.chrom, chromStart + 1, chromEnd - 1)) |
4221664a2bd0
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/translate_bed_sequences commit 2a470e2c775a7427aa530e058510e4dc7b6d8e80"
galaxyp
parents:
0
diff
changeset
|
368 gff_fh.write("%s\t%s\t%s\t%s\t%s\t%s\t%s\t%d\tID=%s\n" % (entry.chrom, 'splice_junc', 'gene', chromStart + 1, chromEnd - 1, entry.score, entry.strand, 0, pep_id)) |
4221664a2bd0
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/translate_bed_sequences commit 2a470e2c775a7427aa530e058510e4dc7b6d8e80"
galaxyp
parents:
0
diff
changeset
|
369 for x in range(blockCount): |
4221664a2bd0
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/translate_bed_sequences commit 2a470e2c775a7427aa530e058510e4dc7b6d8e80"
galaxyp
parents:
0
diff
changeset
|
370 start = chromStart + blockStarts[x] + 1 |
4221664a2bd0
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/translate_bed_sequences commit 2a470e2c775a7427aa530e058510e4dc7b6d8e80"
galaxyp
parents:
0
diff
changeset
|
371 end = start + blockSizes[x] - 1 |
4221664a2bd0
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/translate_bed_sequences commit 2a470e2c775a7427aa530e058510e4dc7b6d8e80"
galaxyp
parents:
0
diff
changeset
|
372 phase = (3 - sum(blockSizes[:x]) % 3) % 3 |
4221664a2bd0
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/translate_bed_sequences commit 2a470e2c775a7427aa530e058510e4dc7b6d8e80"
galaxyp
parents:
0
diff
changeset
|
373 gff_fh.write("%s\t%s\t%s\t%s\t%s\t%s\t%s\t%d\tParent=%s;ID=%s_%d\n" % (entry.chrom, 'splice_junc', 'CDS', start, end, entry.score, entry.strand, phase, pep_id, pep_id, x)) |
4221664a2bd0
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/translate_bed_sequences commit 2a470e2c775a7427aa530e058510e4dc7b6d8e80"
galaxyp
parents:
0
diff
changeset
|
374 # ##gff-version 3 |
4221664a2bd0
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/translate_bed_sequences commit 2a470e2c775a7427aa530e058510e4dc7b6d8e80"
galaxyp
parents:
0
diff
changeset
|
375 # ##sequence-region 19 1 287484 |
4221664a2bd0
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/translate_bed_sequences commit 2a470e2c775a7427aa530e058510e4dc7b6d8e80"
galaxyp
parents:
0
diff
changeset
|
376 # 19 MassSpec peptide 282299 287484 10.0 - 0 ID=TEARLSFYSGHSSFGMYCMVFLALYVQ |
4221664a2bd0
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/translate_bed_sequences commit 2a470e2c775a7427aa530e058510e4dc7b6d8e80"
galaxyp
parents:
0
diff
changeset
|
377 # 19 MassSpec CDS 287474 287484 . - 0 Parent=TEARLSFYSGHSSFGMYCMVFLALYVQ;transcript_id=ENST00000269812 |
4221664a2bd0
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/translate_bed_sequences commit 2a470e2c775a7427aa530e058510e4dc7b6d8e80"
galaxyp
parents:
0
diff
changeset
|
378 # 19 MassSpec CDS 282752 282809 . - 1 Parent=TEARLSFYSGHSSFGMYCMVFLALYVQ;transcript_id=ENST00000269812 |
4221664a2bd0
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/translate_bed_sequences commit 2a470e2c775a7427aa530e058510e4dc7b6d8e80"
galaxyp
parents:
0
diff
changeset
|
379 # 19 MassSpec CDS 282299 282310 . - 0 Parent=TEARLSFYSGHSSFGMYCMVFLALYVQ;transcript_id=ENST00000269812 |
4221664a2bd0
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/translate_bed_sequences commit 2a470e2c775a7427aa530e058510e4dc7b6d8e80"
galaxyp
parents:
0
diff
changeset
|
380 if bed_fh: |
4221664a2bd0
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/translate_bed_sequences commit 2a470e2c775a7427aa530e058510e4dc7b6d8e80"
galaxyp
parents:
0
diff
changeset
|
381 bed_fh.close() |
4221664a2bd0
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/translate_bed_sequences commit 2a470e2c775a7427aa530e058510e4dc7b6d8e80"
galaxyp
parents:
0
diff
changeset
|
382 if gff_fh: |
4221664a2bd0
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/translate_bed_sequences commit 2a470e2c775a7427aa530e058510e4dc7b6d8e80"
galaxyp
parents:
0
diff
changeset
|
383 if gff_fa: |
4221664a2bd0
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/translate_bed_sequences commit 2a470e2c775a7427aa530e058510e4dc7b6d8e80"
galaxyp
parents:
0
diff
changeset
|
384 gff_fa.close() |
4221664a2bd0
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/translate_bed_sequences commit 2a470e2c775a7427aa530e058510e4dc7b6d8e80"
galaxyp
parents:
0
diff
changeset
|
385 else: |
4221664a2bd0
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/translate_bed_sequences commit 2a470e2c775a7427aa530e058510e4dc7b6d8e80"
galaxyp
parents:
0
diff
changeset
|
386 outFile.close() |
4221664a2bd0
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/translate_bed_sequences commit 2a470e2c775a7427aa530e058510e4dc7b6d8e80"
galaxyp
parents:
0
diff
changeset
|
387 gff_fa = open(gff_fa_file, 'r') |
4221664a2bd0
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/translate_bed_sequences commit 2a470e2c775a7427aa530e058510e4dc7b6d8e80"
galaxyp
parents:
0
diff
changeset
|
388 gff_fh.write("##FASTA\n") |
4221664a2bd0
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/translate_bed_sequences commit 2a470e2c775a7427aa530e058510e4dc7b6d8e80"
galaxyp
parents:
0
diff
changeset
|
389 for i, line in enumerate(gff_fa): |
4221664a2bd0
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/translate_bed_sequences commit 2a470e2c775a7427aa530e058510e4dc7b6d8e80"
galaxyp
parents:
0
diff
changeset
|
390 gff_fh.write(line) |
4221664a2bd0
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/translate_bed_sequences commit 2a470e2c775a7427aa530e058510e4dc7b6d8e80"
galaxyp
parents:
0
diff
changeset
|
391 gff_fh.close() |
4221664a2bd0
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/translate_bed_sequences commit 2a470e2c775a7427aa530e058510e4dc7b6d8e80"
galaxyp
parents:
0
diff
changeset
|
392 except Exception as e: |
4221664a2bd0
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/translate_bed_sequences commit 2a470e2c775a7427aa530e058510e4dc7b6d8e80"
galaxyp
parents:
0
diff
changeset
|
393 sys.stderr.write("failed: Error reading %s - %s\n" % (options.input if options.input else 'stdin', e)) |
4221664a2bd0
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/translate_bed_sequences commit 2a470e2c775a7427aa530e058510e4dc7b6d8e80"
galaxyp
parents:
0
diff
changeset
|
394 raise |
4221664a2bd0
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/translate_bed_sequences commit 2a470e2c775a7427aa530e058510e4dc7b6d8e80"
galaxyp
parents:
0
diff
changeset
|
395 exit(1) |
0
d723eb657f1d
planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/translate_bed_sequences commit e04ed4b4960d6109a85c1cc68a2bf4931c8751ef-dirty
galaxyp
parents:
diff
changeset
|
396 |
d723eb657f1d
planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/translate_bed_sequences commit e04ed4b4960d6109a85c1cc68a2bf4931c8751ef-dirty
galaxyp
parents:
diff
changeset
|
397 |
2
4221664a2bd0
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/translate_bed_sequences commit 2a470e2c775a7427aa530e058510e4dc7b6d8e80"
galaxyp
parents:
0
diff
changeset
|
398 if __name__ == "__main__": |
4221664a2bd0
"planemo upload for repository https://github.com/galaxyproteomics/tools-galaxyp/tree/master/tools/translate_bed_sequences commit 2a470e2c775a7427aa530e058510e4dc7b6d8e80"
galaxyp
parents:
0
diff
changeset
|
399 __main__() |