Mercurial > repos > iuc > ngsutils_bam_filter

annotate ngsutils/support/regions.py @ 2:7a68005de299 draft

Find changesets by keywords (author, files, the commit message), revision number or hash, or revset expression.
planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/ngsutils commit 9a243c616a4a3156347e38fdb5f35863ae5133f9
author iuc
date Sun, 27 Nov 2016 15:01:21 -0500
parents 4e4e4093d65d
children
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
rev   line source
0
4e4e4093d65d planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/ngsutils commit 09194687c74a424732f8b0c017cbb942aad89068
iuc
parents:
diff changeset
1 class RangeMatch(object):
4e4e4093d65d planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/ngsutils commit 09194687c74a424732f8b0c017cbb942aad89068
iuc
parents:
diff changeset
2 '''
4e4e4093d65d planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/ngsutils commit 09194687c74a424732f8b0c017cbb942aad89068
iuc
parents:
diff changeset
3 Simple genomic ranges. You can define chrom:start-end ranges, then ask if a
4e4e4093d65d planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/ngsutils commit 09194687c74a424732f8b0c017cbb942aad89068
iuc
parents:
diff changeset
4 particular genomic coordinate maps to any of those ranges. This is less-
4e4e4093d65d planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/ngsutils commit 09194687c74a424732f8b0c017cbb942aad89068
iuc
parents:
diff changeset
5 efficient than an R-Tree, but easier to code.
4e4e4093d65d planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/ngsutils commit 09194687c74a424732f8b0c017cbb942aad89068
iuc
parents:
diff changeset
6 '''
2
7a68005de299 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/ngsutils commit 9a243c616a4a3156347e38fdb5f35863ae5133f9
iuc
parents: 0
diff changeset
7
0
4e4e4093d65d planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/ngsutils commit 09194687c74a424732f8b0c017cbb942aad89068
iuc
parents:
diff changeset
8 def __init__(self, name):
4e4e4093d65d planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/ngsutils commit 09194687c74a424732f8b0c017cbb942aad89068
iuc
parents:
diff changeset
9 self.ranges = {}
4e4e4093d65d planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/ngsutils commit 09194687c74a424732f8b0c017cbb942aad89068
iuc
parents:
diff changeset
10 self.name = name
4e4e4093d65d planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/ngsutils commit 09194687c74a424732f8b0c017cbb942aad89068
iuc
parents:
diff changeset
11
4e4e4093d65d planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/ngsutils commit 09194687c74a424732f8b0c017cbb942aad89068
iuc
parents:
diff changeset
12 def add_range(self, chrom, strand, start, end):
2
7a68005de299 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/ngsutils commit 9a243c616a4a3156347e38fdb5f35863ae5133f9
iuc
parents: 0
diff changeset
13 if chrom not in self.ranges:
0
4e4e4093d65d planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/ngsutils commit 09194687c74a424732f8b0c017cbb942aad89068
iuc
parents:
diff changeset
14 self.ranges[chrom] = {}
4e4e4093d65d planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/ngsutils commit 09194687c74a424732f8b0c017cbb942aad89068
iuc
parents:
diff changeset
15
4e4e4093d65d planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/ngsutils commit 09194687c74a424732f8b0c017cbb942aad89068
iuc
parents:
diff changeset
16 bin = start / 100000
2
7a68005de299 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/ngsutils commit 9a243c616a4a3156347e38fdb5f35863ae5133f9
iuc
parents: 0
diff changeset
17 if bin not in self.ranges[chrom]:
0
4e4e4093d65d planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/ngsutils commit 09194687c74a424732f8b0c017cbb942aad89068
iuc
parents:
diff changeset
18 self.ranges[chrom][bin] = []
4e4e4093d65d planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/ngsutils commit 09194687c74a424732f8b0c017cbb942aad89068
iuc
parents:
diff changeset
19 self.ranges[chrom][bin].insert(0, (start, end, strand))
4e4e4093d65d planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/ngsutils commit 09194687c74a424732f8b0c017cbb942aad89068
iuc
parents:
diff changeset
20
4e4e4093d65d planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/ngsutils commit 09194687c74a424732f8b0c017cbb942aad89068
iuc
parents:
diff changeset
21 if (end / 100000) != bin:
4e4e4093d65d planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/ngsutils commit 09194687c74a424732f8b0c017cbb942aad89068
iuc
parents:
diff changeset
22 for bin in xrange(bin + 1, (end / 100000) + 1):
2
7a68005de299 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/ngsutils commit 9a243c616a4a3156347e38fdb5f35863ae5133f9
iuc
parents: 0
diff changeset
23 if bin not in self.ranges[chrom]:
0
4e4e4093d65d planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/ngsutils commit 09194687c74a424732f8b0c017cbb942aad89068
iuc
parents:
diff changeset
24 self.ranges[chrom][bin] = []
4e4e4093d65d planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/ngsutils commit 09194687c74a424732f8b0c017cbb942aad89068
iuc
parents:
diff changeset
25 self.ranges[chrom][bin].insert(0, (start, end, strand))
4e4e4093d65d planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/ngsutils commit 09194687c74a424732f8b0c017cbb942aad89068
iuc
parents:
diff changeset
26
4e4e4093d65d planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/ngsutils commit 09194687c74a424732f8b0c017cbb942aad89068
iuc
parents:
diff changeset
27 def get_tag(self, chrom, strand, pos, ignore_strand=False):
4e4e4093d65d planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/ngsutils commit 09194687c74a424732f8b0c017cbb942aad89068
iuc
parents:
diff changeset
28 '''
4e4e4093d65d planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/ngsutils commit 09194687c74a424732f8b0c017cbb942aad89068
iuc
parents:
diff changeset
29 returns (region, is_reverse_orientation)
4e4e4093d65d planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/ngsutils commit 09194687c74a424732f8b0c017cbb942aad89068
iuc
parents:
diff changeset
30 '''
2
7a68005de299 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/ngsutils commit 9a243c616a4a3156347e38fdb5f35863ae5133f9
iuc
parents: 0
diff changeset
31 if chrom not in self.ranges:
0
4e4e4093d65d planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/ngsutils commit 09194687c74a424732f8b0c017cbb942aad89068
iuc
parents:
diff changeset
32 return None, False
4e4e4093d65d planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/ngsutils commit 09194687c74a424732f8b0c017cbb942aad89068
iuc
parents:
diff changeset
33 bin = pos / 100000
2
7a68005de299 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/ngsutils commit 9a243c616a4a3156347e38fdb5f35863ae5133f9
iuc
parents: 0
diff changeset
34 if bin not in self.ranges[chrom]:
0
4e4e4093d65d planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/ngsutils commit 09194687c74a424732f8b0c017cbb942aad89068
iuc
parents:
diff changeset
35 return None, False
4e4e4093d65d planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/ngsutils commit 09194687c74a424732f8b0c017cbb942aad89068
iuc
parents:
diff changeset
36 for start, end, r_strand in self.ranges[chrom][bin]:
4e4e4093d65d planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/ngsutils commit 09194687c74a424732f8b0c017cbb942aad89068
iuc
parents:
diff changeset
37 if pos >= start and pos <= end:
4e4e4093d65d planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/ngsutils commit 09194687c74a424732f8b0c017cbb942aad89068
iuc
parents:
diff changeset
38 if ignore_strand or strand == r_strand:
4e4e4093d65d planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/ngsutils commit 09194687c74a424732f8b0c017cbb942aad89068
iuc
parents:
diff changeset
39 return self.name, False
4e4e4093d65d planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/ngsutils commit 09194687c74a424732f8b0c017cbb942aad89068
iuc
parents:
diff changeset
40 return self.name, True
4e4e4093d65d planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/ngsutils commit 09194687c74a424732f8b0c017cbb942aad89068
iuc
parents:
diff changeset
41 return None, False
4e4e4093d65d planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/ngsutils commit 09194687c74a424732f8b0c017cbb942aad89068
iuc
parents:
diff changeset
42
4e4e4093d65d planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/ngsutils commit 09194687c74a424732f8b0c017cbb942aad89068
iuc
parents:
diff changeset
43
4e4e4093d65d planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/ngsutils commit 09194687c74a424732f8b0c017cbb942aad89068
iuc
parents:
diff changeset
44 class RegionTagger(object):
4e4e4093d65d planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/ngsutils commit 09194687c74a424732f8b0c017cbb942aad89068
iuc
parents:
diff changeset
45 def __init__(self, gtf, valid_chroms=None, only_first_fragment=True):
4e4e4093d65d planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/ngsutils commit 09194687c74a424732f8b0c017cbb942aad89068
iuc
parents:
diff changeset
46 self.regions = []
4e4e4093d65d planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/ngsutils commit 09194687c74a424732f8b0c017cbb942aad89068
iuc
parents:
diff changeset
47 self.counts = {}
4e4e4093d65d planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/ngsutils commit 09194687c74a424732f8b0c017cbb942aad89068
iuc
parents:
diff changeset
48 self.only_first_fragment = only_first_fragment
4e4e4093d65d planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/ngsutils commit 09194687c74a424732f8b0c017cbb942aad89068
iuc
parents:
diff changeset
49
4e4e4093d65d planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/ngsutils commit 09194687c74a424732f8b0c017cbb942aad89068
iuc
parents:
diff changeset
50 coding = RangeMatch('coding')
4e4e4093d65d planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/ngsutils commit 09194687c74a424732f8b0c017cbb942aad89068
iuc
parents:
diff changeset
51 exons = RangeMatch('other-exon')
4e4e4093d65d planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/ngsutils commit 09194687c74a424732f8b0c017cbb942aad89068
iuc
parents:
diff changeset
52 utr_5 = RangeMatch('utr-5')
4e4e4093d65d planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/ngsutils commit 09194687c74a424732f8b0c017cbb942aad89068
iuc
parents:
diff changeset
53 utr_3 = RangeMatch('utr-3')
4e4e4093d65d planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/ngsutils commit 09194687c74a424732f8b0c017cbb942aad89068
iuc
parents:
diff changeset
54 introns = RangeMatch('intron')
4e4e4093d65d planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/ngsutils commit 09194687c74a424732f8b0c017cbb942aad89068
iuc
parents:
diff changeset
55 promoters = RangeMatch('promoter')
4e4e4093d65d planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/ngsutils commit 09194687c74a424732f8b0c017cbb942aad89068
iuc
parents:
diff changeset
56
4e4e4093d65d planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/ngsutils commit 09194687c74a424732f8b0c017cbb942aad89068
iuc
parents:
diff changeset
57 for gene in gtf.genes:
2
7a68005de299 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/ngsutils commit 9a243c616a4a3156347e38fdb5f35863ae5133f9
iuc
parents: 0
diff changeset
58 if valid_chroms and gene.chrom not in valid_chroms:
0
4e4e4093d65d planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/ngsutils commit 09194687c74a424732f8b0c017cbb942aad89068
iuc
parents:
diff changeset
59 continue
4e4e4093d65d planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/ngsutils commit 09194687c74a424732f8b0c017cbb942aad89068
iuc
parents:
diff changeset
60 if gene.strand == '+':
4e4e4093d65d planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/ngsutils commit 09194687c74a424732f8b0c017cbb942aad89068
iuc
parents:
diff changeset
61 promoters.add_range(gene.chrom, gene.strand, gene.start - 2000, gene.start)
4e4e4093d65d planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/ngsutils commit 09194687c74a424732f8b0c017cbb942aad89068
iuc
parents:
diff changeset
62 else:
4e4e4093d65d planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/ngsutils commit 09194687c74a424732f8b0c017cbb942aad89068
iuc
parents:
diff changeset
63 promoters.add_range(gene.chrom, gene.strand, gene.end, gene.end + 2000)
4e4e4093d65d planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/ngsutils commit 09194687c74a424732f8b0c017cbb942aad89068
iuc
parents:
diff changeset
64
4e4e4093d65d planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/ngsutils commit 09194687c74a424732f8b0c017cbb942aad89068
iuc
parents:
diff changeset
65 for transcript in gene.transcripts:
4e4e4093d65d planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/ngsutils commit 09194687c74a424732f8b0c017cbb942aad89068
iuc
parents:
diff changeset
66 if transcript.has_cds:
4e4e4093d65d planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/ngsutils commit 09194687c74a424732f8b0c017cbb942aad89068
iuc
parents:
diff changeset
67 for start, end in transcript.cds:
4e4e4093d65d planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/ngsutils commit 09194687c74a424732f8b0c017cbb942aad89068
iuc
parents:
diff changeset
68 coding.add_range(gene.chrom, gene.strand, start, end)
4e4e4093d65d planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/ngsutils commit 09194687c74a424732f8b0c017cbb942aad89068
iuc
parents:
diff changeset
69
4e4e4093d65d planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/ngsutils commit 09194687c74a424732f8b0c017cbb942aad89068
iuc
parents:
diff changeset
70 # TODO: Fix this so that it iterates over exons in the 5'/3' UTRS
4e4e4093d65d planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/ngsutils commit 09194687c74a424732f8b0c017cbb942aad89068
iuc
parents:
diff changeset
71 for s, e in transcript.utr_5:
4e4e4093d65d planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/ngsutils commit 09194687c74a424732f8b0c017cbb942aad89068
iuc
parents:
diff changeset
72 utr_5.add_range(gene.chrom, gene.strand, s, e)
4e4e4093d65d planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/ngsutils commit 09194687c74a424732f8b0c017cbb942aad89068
iuc
parents:
diff changeset
73 for s, e in transcript.utr_3:
4e4e4093d65d planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/ngsutils commit 09194687c74a424732f8b0c017cbb942aad89068
iuc
parents:
diff changeset
74 utr_3.add_range(gene.chrom, gene.strand, s, e)
4e4e4093d65d planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/ngsutils commit 09194687c74a424732f8b0c017cbb942aad89068
iuc
parents:
diff changeset
75
4e4e4093d65d planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/ngsutils commit 09194687c74a424732f8b0c017cbb942aad89068
iuc
parents:
diff changeset
76 last_end = None
4e4e4093d65d planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/ngsutils commit 09194687c74a424732f8b0c017cbb942aad89068
iuc
parents:
diff changeset
77 for start, end in transcript.exons:
4e4e4093d65d planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/ngsutils commit 09194687c74a424732f8b0c017cbb942aad89068
iuc
parents:
diff changeset
78 if last_end:
4e4e4093d65d planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/ngsutils commit 09194687c74a424732f8b0c017cbb942aad89068
iuc
parents:
diff changeset
79 introns.add_range(gene.chrom, gene.strand, last_end, start)
4e4e4093d65d planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/ngsutils commit 09194687c74a424732f8b0c017cbb942aad89068
iuc
parents:
diff changeset
80 exons.add_range(gene.chrom, gene.strand, start, end)
4e4e4093d65d planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/ngsutils commit 09194687c74a424732f8b0c017cbb942aad89068
iuc
parents:
diff changeset
81 last_end = end
4e4e4093d65d planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/ngsutils commit 09194687c74a424732f8b0c017cbb942aad89068
iuc
parents:
diff changeset
82
4e4e4093d65d planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/ngsutils commit 09194687c74a424732f8b0c017cbb942aad89068
iuc
parents:
diff changeset
83 self.regions.append(coding)
4e4e4093d65d planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/ngsutils commit 09194687c74a424732f8b0c017cbb942aad89068
iuc
parents:
diff changeset
84 self.regions.append(utr_5)
4e4e4093d65d planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/ngsutils commit 09194687c74a424732f8b0c017cbb942aad89068
iuc
parents:
diff changeset
85 self.regions.append(utr_3)
4e4e4093d65d planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/ngsutils commit 09194687c74a424732f8b0c017cbb942aad89068
iuc
parents:
diff changeset
86 self.regions.append(exons)
4e4e4093d65d planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/ngsutils commit 09194687c74a424732f8b0c017cbb942aad89068
iuc
parents:
diff changeset
87 self.regions.append(introns)
4e4e4093d65d planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/ngsutils commit 09194687c74a424732f8b0c017cbb942aad89068
iuc
parents:
diff changeset
88 self.regions.append(promoters)
4e4e4093d65d planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/ngsutils commit 09194687c74a424732f8b0c017cbb942aad89068
iuc
parents:
diff changeset
89
4e4e4093d65d planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/ngsutils commit 09194687c74a424732f8b0c017cbb942aad89068
iuc
parents:
diff changeset
90 self.counts['coding'] = 0
4e4e4093d65d planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/ngsutils commit 09194687c74a424732f8b0c017cbb942aad89068
iuc
parents:
diff changeset
91 self.counts['coding-rev'] = 0
4e4e4093d65d planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/ngsutils commit 09194687c74a424732f8b0c017cbb942aad89068
iuc
parents:
diff changeset
92 self.counts['other-exon'] = 0
4e4e4093d65d planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/ngsutils commit 09194687c74a424732f8b0c017cbb942aad89068
iuc
parents:
diff changeset
93 self.counts['utr-5'] = 0
4e4e4093d65d planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/ngsutils commit 09194687c74a424732f8b0c017cbb942aad89068
iuc
parents:
diff changeset
94 self.counts['utr-3'] = 0
4e4e4093d65d planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/ngsutils commit 09194687c74a424732f8b0c017cbb942aad89068
iuc
parents:
diff changeset
95 self.counts['utr-5-rev'] = 0
4e4e4093d65d planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/ngsutils commit 09194687c74a424732f8b0c017cbb942aad89068
iuc
parents:
diff changeset
96 self.counts['utr-3-rev'] = 0
4e4e4093d65d planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/ngsutils commit 09194687c74a424732f8b0c017cbb942aad89068
iuc
parents:
diff changeset
97 self.counts['intron'] = 0
4e4e4093d65d planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/ngsutils commit 09194687c74a424732f8b0c017cbb942aad89068
iuc
parents:
diff changeset
98 self.counts['promoter'] = 0
4e4e4093d65d planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/ngsutils commit 09194687c74a424732f8b0c017cbb942aad89068
iuc
parents:
diff changeset
99 self.counts['other-exon-rev'] = 0
4e4e4093d65d planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/ngsutils commit 09194687c74a424732f8b0c017cbb942aad89068
iuc
parents:
diff changeset
100 self.counts['intron-rev'] = 0
4e4e4093d65d planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/ngsutils commit 09194687c74a424732f8b0c017cbb942aad89068
iuc
parents:
diff changeset
101 self.counts['promoter-rev'] = 0
4e4e4093d65d planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/ngsutils commit 09194687c74a424732f8b0c017cbb942aad89068
iuc
parents:
diff changeset
102 self.counts['junction'] = 0
4e4e4093d65d planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/ngsutils commit 09194687c74a424732f8b0c017cbb942aad89068
iuc
parents:
diff changeset
103 self.counts['intergenic'] = 0
4e4e4093d65d planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/ngsutils commit 09194687c74a424732f8b0c017cbb942aad89068
iuc
parents:
diff changeset
104 self.counts['mitochondrial'] = 0
4e4e4093d65d planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/ngsutils commit 09194687c74a424732f8b0c017cbb942aad89068
iuc
parents:
diff changeset
105
4e4e4093d65d planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/ngsutils commit 09194687c74a424732f8b0c017cbb942aad89068
iuc
parents:
diff changeset
106 def add_read(self, read, chrom):
4e4e4093d65d planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/ngsutils commit 09194687c74a424732f8b0c017cbb942aad89068
iuc
parents:
diff changeset
107 if read.is_unmapped:
4e4e4093d65d planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/ngsutils commit 09194687c74a424732f8b0c017cbb942aad89068
iuc
parents:
diff changeset
108 return
2
7a68005de299 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/ngsutils commit 9a243c616a4a3156347e38fdb5f35863ae5133f9
iuc
parents: 0
diff changeset
109
0
4e4e4093d65d planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/ngsutils commit 09194687c74a424732f8b0c017cbb942aad89068
iuc
parents:
diff changeset
110 if self.only_first_fragment and read.is_paired and not read.is_read1:
4e4e4093d65d planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/ngsutils commit 09194687c74a424732f8b0c017cbb942aad89068
iuc
parents:
diff changeset
111 return
4e4e4093d65d planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/ngsutils commit 09194687c74a424732f8b0c017cbb942aad89068
iuc
parents:
diff changeset
112
4e4e4093d65d planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/ngsutils commit 09194687c74a424732f8b0c017cbb942aad89068
iuc
parents:
diff changeset
113 tag = None
4e4e4093d65d planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/ngsutils commit 09194687c74a424732f8b0c017cbb942aad89068
iuc
parents:
diff changeset
114 is_rev = False
4e4e4093d65d planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/ngsutils commit 09194687c74a424732f8b0c017cbb942aad89068
iuc
parents:
diff changeset
115
4e4e4093d65d planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/ngsutils commit 09194687c74a424732f8b0c017cbb942aad89068
iuc
parents:
diff changeset
116 strand = '-' if read.is_reverse else '+'
4e4e4093d65d planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/ngsutils commit 09194687c74a424732f8b0c017cbb942aad89068
iuc
parents:
diff changeset
117
4e4e4093d65d planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/ngsutils commit 09194687c74a424732f8b0c017cbb942aad89068
iuc
parents:
diff changeset
118 if chrom == 'chrM':
4e4e4093d65d planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/ngsutils commit 09194687c74a424732f8b0c017cbb942aad89068
iuc
parents:
diff changeset
119 tag = 'mitochondrial'
4e4e4093d65d planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/ngsutils commit 09194687c74a424732f8b0c017cbb942aad89068
iuc
parents:
diff changeset
120
4e4e4093d65d planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/ngsutils commit 09194687c74a424732f8b0c017cbb942aad89068
iuc
parents:
diff changeset
121 if not tag:
4e4e4093d65d planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/ngsutils commit 09194687c74a424732f8b0c017cbb942aad89068
iuc
parents:
diff changeset
122 for op, length in read.cigar:
4e4e4093d65d planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/ngsutils commit 09194687c74a424732f8b0c017cbb942aad89068
iuc
parents:
diff changeset
123 if op == 3:
4e4e4093d65d planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/ngsutils commit 09194687c74a424732f8b0c017cbb942aad89068
iuc
parents:
diff changeset
124 tag = 'junction'
4e4e4093d65d planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/ngsutils commit 09194687c74a424732f8b0c017cbb942aad89068
iuc
parents:
diff changeset
125 break
4e4e4093d65d planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/ngsutils commit 09194687c74a424732f8b0c017cbb942aad89068
iuc
parents:
diff changeset
126
4e4e4093d65d planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/ngsutils commit 09194687c74a424732f8b0c017cbb942aad89068
iuc
parents:
diff changeset
127 if not tag:
4e4e4093d65d planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/ngsutils commit 09194687c74a424732f8b0c017cbb942aad89068
iuc
parents:
diff changeset
128 for region in self.regions:
4e4e4093d65d planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/ngsutils commit 09194687c74a424732f8b0c017cbb942aad89068
iuc
parents:
diff changeset
129 tag, is_rev = region.get_tag(chrom, strand, read.pos)
4e4e4093d65d planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/ngsutils commit 09194687c74a424732f8b0c017cbb942aad89068
iuc
parents:
diff changeset
130 if tag:
4e4e4093d65d planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/ngsutils commit 09194687c74a424732f8b0c017cbb942aad89068
iuc
parents:
diff changeset
131 break
4e4e4093d65d planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/ngsutils commit 09194687c74a424732f8b0c017cbb942aad89068
iuc
parents:
diff changeset
132
4e4e4093d65d planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/ngsutils commit 09194687c74a424732f8b0c017cbb942aad89068
iuc
parents:
diff changeset
133 if not tag:
4e4e4093d65d planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/ngsutils commit 09194687c74a424732f8b0c017cbb942aad89068
iuc
parents:
diff changeset
134 tag = 'intergenic'
4e4e4093d65d planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/ngsutils commit 09194687c74a424732f8b0c017cbb942aad89068
iuc
parents:
diff changeset
135
4e4e4093d65d planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/ngsutils commit 09194687c74a424732f8b0c017cbb942aad89068
iuc
parents:
diff changeset
136 if tag:
4e4e4093d65d planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/ngsutils commit 09194687c74a424732f8b0c017cbb942aad89068
iuc
parents:
diff changeset
137 if is_rev:
4e4e4093d65d planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/ngsutils commit 09194687c74a424732f8b0c017cbb942aad89068
iuc
parents:
diff changeset
138 self.counts['%s-rev' % tag] += 1
4e4e4093d65d planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/ngsutils commit 09194687c74a424732f8b0c017cbb942aad89068
iuc
parents:
diff changeset
139 else:
4e4e4093d65d planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/ngsutils commit 09194687c74a424732f8b0c017cbb942aad89068
iuc
parents:
diff changeset
140 self.counts[tag] += 1
4e4e4093d65d planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/ngsutils commit 09194687c74a424732f8b0c017cbb942aad89068
iuc
parents:
diff changeset
141
4e4e4093d65d planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/ngsutils commit 09194687c74a424732f8b0c017cbb942aad89068
iuc
parents:
diff changeset
142 return tag
4e4e4093d65d planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/ngsutils commit 09194687c74a424732f8b0c017cbb942aad89068
iuc
parents:
diff changeset
143
4e4e4093d65d planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/ngsutils commit 09194687c74a424732f8b0c017cbb942aad89068
iuc
parents:
diff changeset
144 def tag_region(self, chrom, start, end, strand):
4e4e4093d65d planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/ngsutils commit 09194687c74a424732f8b0c017cbb942aad89068
iuc
parents:
diff changeset
145 tag = None
4e4e4093d65d planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/ngsutils commit 09194687c74a424732f8b0c017cbb942aad89068
iuc
parents:
diff changeset
146 is_rev = False
4e4e4093d65d planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/ngsutils commit 09194687c74a424732f8b0c017cbb942aad89068
iuc
parents:
diff changeset
147
4e4e4093d65d planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/ngsutils commit 09194687c74a424732f8b0c017cbb942aad89068
iuc
parents:
diff changeset
148 if chrom == 'chrM' or chrom == 'M':
4e4e4093d65d planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/ngsutils commit 09194687c74a424732f8b0c017cbb942aad89068
iuc
parents:
diff changeset
149 tag = 'mitochondrial'
4e4e4093d65d planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/ngsutils commit 09194687c74a424732f8b0c017cbb942aad89068
iuc
parents:
diff changeset
150
4e4e4093d65d planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/ngsutils commit 09194687c74a424732f8b0c017cbb942aad89068
iuc
parents:
diff changeset
151 if not tag:
4e4e4093d65d planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/ngsutils commit 09194687c74a424732f8b0c017cbb942aad89068
iuc
parents:
diff changeset
152 for region in self.regions:
4e4e4093d65d planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/ngsutils commit 09194687c74a424732f8b0c017cbb942aad89068
iuc
parents:
diff changeset
153 tag, is_rev = region.get_tag(chrom, strand, start)
4e4e4093d65d planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/ngsutils commit 09194687c74a424732f8b0c017cbb942aad89068
iuc
parents:
diff changeset
154 if is_rev:
4e4e4093d65d planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/ngsutils commit 09194687c74a424732f8b0c017cbb942aad89068
iuc
parents:
diff changeset
155 tag = '%s-rev' % tag
4e4e4093d65d planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/ngsutils commit 09194687c74a424732f8b0c017cbb942aad89068
iuc
parents:
diff changeset
156
4e4e4093d65d planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/ngsutils commit 09194687c74a424732f8b0c017cbb942aad89068
iuc
parents:
diff changeset
157 if start != end:
4e4e4093d65d planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/ngsutils commit 09194687c74a424732f8b0c017cbb942aad89068
iuc
parents:
diff changeset
158 endtag, is_rev = region.get_tag(chrom, strand, end)
4e4e4093d65d planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/ngsutils commit 09194687c74a424732f8b0c017cbb942aad89068
iuc
parents:
diff changeset
159 if is_rev:
4e4e4093d65d planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/ngsutils commit 09194687c74a424732f8b0c017cbb942aad89068
iuc
parents:
diff changeset
160 endtag = '%s-rev' % endtag
4e4e4093d65d planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/ngsutils commit 09194687c74a424732f8b0c017cbb942aad89068
iuc
parents:
diff changeset
161
4e4e4093d65d planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/ngsutils commit 09194687c74a424732f8b0c017cbb942aad89068
iuc
parents:
diff changeset
162 if tag and endtag and endtag != tag:
4e4e4093d65d planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/ngsutils commit 09194687c74a424732f8b0c017cbb942aad89068
iuc
parents:
diff changeset
163 tag = '%s/%s' % (tag, endtag)
4e4e4093d65d planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/ngsutils commit 09194687c74a424732f8b0c017cbb942aad89068
iuc
parents:
diff changeset
164
4e4e4093d65d planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/ngsutils commit 09194687c74a424732f8b0c017cbb942aad89068
iuc
parents:
diff changeset
165 if not tag:
2
7a68005de299 planemo upload for repository https://github.com/galaxyproject/tools-iuc/tree/master/tools/ngsutils commit 9a243c616a4a3156347e38fdb5f35863ae5133f9
iuc
parents: 0
diff changeset
166 tag = 'intergenic'