Mercurial > repos > artbio > small_rna_signatures

annotate overlapping_reads.py @ 4:20d28cfdeefe draft

Find changesets by keywords (author, files, the commit message), revision number or hash, or revset expression.
planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_signatures commit cfdc08418887bfe4a35588cd78d0a2b6ffa6e19e
author artbio
date Fri, 08 Sep 2017 04:44:22 -0400
parents 4d9682bd3a6b
children a7fd04208764
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
rev   line source
1
6f1378738798 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_signatures commit 6133bb114c76a795fa12a4a11edb1a8b80fd104d
artbio
parents:
diff changeset
1 import argparse
6f1378738798 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_signatures commit 6133bb114c76a795fa12a4a11edb1a8b80fd104d
artbio
parents:
diff changeset
2 from collections import defaultdict
6f1378738798 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_signatures commit 6133bb114c76a795fa12a4a11edb1a8b80fd104d
artbio
parents:
diff changeset
3
6f1378738798 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_signatures commit 6133bb114c76a795fa12a4a11edb1a8b80fd104d
artbio
parents:
diff changeset
4 import pysam
6f1378738798 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_signatures commit 6133bb114c76a795fa12a4a11edb1a8b80fd104d
artbio
parents:
diff changeset
5
6f1378738798 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_signatures commit 6133bb114c76a795fa12a4a11edb1a8b80fd104d
artbio
parents:
diff changeset
6
6f1378738798 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_signatures commit 6133bb114c76a795fa12a4a11edb1a8b80fd104d
artbio
parents:
diff changeset
7 def Parser():
6f1378738798 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_signatures commit 6133bb114c76a795fa12a4a11edb1a8b80fd104d
artbio
parents:
diff changeset
8 the_parser = argparse.ArgumentParser()
6f1378738798 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_signatures commit 6133bb114c76a795fa12a4a11edb1a8b80fd104d
artbio
parents:
diff changeset
9 the_parser.add_argument(
6f1378738798 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_signatures commit 6133bb114c76a795fa12a4a11edb1a8b80fd104d
artbio
parents:
diff changeset
10 '--input', action="store", type=str, help="bam alignment file")
6f1378738798 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_signatures commit 6133bb114c76a795fa12a4a11edb1a8b80fd104d
artbio
parents:
diff changeset
11 the_parser.add_argument(
6f1378738798 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_signatures commit 6133bb114c76a795fa12a4a11edb1a8b80fd104d
artbio
parents:
diff changeset
12 '--minquery', type=int,
6f1378738798 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_signatures commit 6133bb114c76a795fa12a4a11edb1a8b80fd104d
artbio
parents:
diff changeset
13 help="Minimum readsize of query reads (nt) - must be an integer")
6f1378738798 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_signatures commit 6133bb114c76a795fa12a4a11edb1a8b80fd104d
artbio
parents:
diff changeset
14 the_parser.add_argument(
6f1378738798 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_signatures commit 6133bb114c76a795fa12a4a11edb1a8b80fd104d
artbio
parents:
diff changeset
15 '--maxquery', type=int,
6f1378738798 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_signatures commit 6133bb114c76a795fa12a4a11edb1a8b80fd104d
artbio
parents:
diff changeset
16 help="Maximum readsize of query reads (nt) - must be an integer")
6f1378738798 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_signatures commit 6133bb114c76a795fa12a4a11edb1a8b80fd104d
artbio
parents:
diff changeset
17 the_parser.add_argument(
6f1378738798 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_signatures commit 6133bb114c76a795fa12a4a11edb1a8b80fd104d
artbio
parents:
diff changeset
18 '--mintarget', type=int,
6f1378738798 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_signatures commit 6133bb114c76a795fa12a4a11edb1a8b80fd104d
artbio
parents:
diff changeset
19 help="Minimum readsize of target reads (nt) - must be an integer")
6f1378738798 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_signatures commit 6133bb114c76a795fa12a4a11edb1a8b80fd104d
artbio
parents:
diff changeset
20 the_parser.add_argument(
6f1378738798 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_signatures commit 6133bb114c76a795fa12a4a11edb1a8b80fd104d
artbio
parents:
diff changeset
21 '--maxtarget', type=int,
6f1378738798 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_signatures commit 6133bb114c76a795fa12a4a11edb1a8b80fd104d
artbio
parents:
diff changeset
22 help="Maximum readsize of target reads (nt) - must be an integer")
6f1378738798 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_signatures commit 6133bb114c76a795fa12a4a11edb1a8b80fd104d
artbio
parents:
diff changeset
23 the_parser.add_argument(
6f1378738798 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_signatures commit 6133bb114c76a795fa12a4a11edb1a8b80fd104d
artbio
parents:
diff changeset
24 '--overlap', type=int,
6f1378738798 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_signatures commit 6133bb114c76a795fa12a4a11edb1a8b80fd104d
artbio
parents:
diff changeset
25 help="Overlap analyzed (nt) - must be an integer")
6f1378738798 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_signatures commit 6133bb114c76a795fa12a4a11edb1a8b80fd104d
artbio
parents:
diff changeset
26 the_parser.add_argument(
6f1378738798 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_signatures commit 6133bb114c76a795fa12a4a11edb1a8b80fd104d
artbio
parents:
diff changeset
27 '--output', action="store", type=str,
6f1378738798 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_signatures commit 6133bb114c76a795fa12a4a11edb1a8b80fd104d
artbio
parents:
diff changeset
28 help="Pairable sequences")
6f1378738798 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_signatures commit 6133bb114c76a795fa12a4a11edb1a8b80fd104d
artbio
parents:
diff changeset
29 args = the_parser.parse_args()
6f1378738798 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_signatures commit 6133bb114c76a795fa12a4a11edb1a8b80fd104d
artbio
parents:
diff changeset
30 return args
6f1378738798 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_signatures commit 6133bb114c76a795fa12a4a11edb1a8b80fd104d
artbio
parents:
diff changeset
31
6f1378738798 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_signatures commit 6133bb114c76a795fa12a4a11edb1a8b80fd104d
artbio
parents:
diff changeset
32
6f1378738798 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_signatures commit 6133bb114c76a795fa12a4a11edb1a8b80fd104d
artbio
parents:
diff changeset
33 class Map:
6f1378738798 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_signatures commit 6133bb114c76a795fa12a4a11edb1a8b80fd104d
artbio
parents:
diff changeset
34
6f1378738798 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_signatures commit 6133bb114c76a795fa12a4a11edb1a8b80fd104d
artbio
parents:
diff changeset
35 def __init__(self, bam_file):
6f1378738798 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_signatures commit 6133bb114c76a795fa12a4a11edb1a8b80fd104d
artbio
parents:
diff changeset
36 self.bam_object = pysam.AlignmentFile(bam_file, 'rb')
6f1378738798 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_signatures commit 6133bb114c76a795fa12a4a11edb1a8b80fd104d
artbio
parents:
diff changeset
37 self.chromosomes = dict(zip(self.bam_object.references,
6f1378738798 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_signatures commit 6133bb114c76a795fa12a4a11edb1a8b80fd104d
artbio
parents:
diff changeset
38 self.bam_object.lengths))
3
4d9682bd3a6b planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_signatures commit 96ed5824190aff281cc3aa47dc60fc66aac41db3
artbio
parents: 1
diff changeset
39 self.all_query_positions = self.query_positions(self.bam_object)
4d9682bd3a6b planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_signatures commit 96ed5824190aff281cc3aa47dc60fc66aac41db3
artbio
parents: 1
diff changeset
40 self.readdic = self.make_readdic(self.bam_object)
1
6f1378738798 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_signatures commit 6133bb114c76a795fa12a4a11edb1a8b80fd104d
artbio
parents:
diff changeset
41
3
4d9682bd3a6b planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_signatures commit 96ed5824190aff281cc3aa47dc60fc66aac41db3
artbio
parents: 1
diff changeset
42 def make_readdic(self, bam_object):
4d9682bd3a6b planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_signatures commit 96ed5824190aff281cc3aa47dc60fc66aac41db3
artbio
parents: 1
diff changeset
43 readdic = defaultdict(int)
4d9682bd3a6b planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_signatures commit 96ed5824190aff281cc3aa47dc60fc66aac41db3
artbio
parents: 1
diff changeset
44 for read in bam_object.fetch():
4d9682bd3a6b planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_signatures commit 96ed5824190aff281cc3aa47dc60fc66aac41db3
artbio
parents: 1
diff changeset
45 readdic[read.query_sequence] += 1
4d9682bd3a6b planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_signatures commit 96ed5824190aff281cc3aa47dc60fc66aac41db3
artbio
parents: 1
diff changeset
46 return readdic
4d9682bd3a6b planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_signatures commit 96ed5824190aff281cc3aa47dc60fc66aac41db3
artbio
parents: 1
diff changeset
47
4d9682bd3a6b planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_signatures commit 96ed5824190aff281cc3aa47dc60fc66aac41db3
artbio
parents: 1
diff changeset
48 def query_positions(self, bam_object):
4d9682bd3a6b planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_signatures commit 96ed5824190aff281cc3aa47dc60fc66aac41db3
artbio
parents: 1
diff changeset
49 all_query_positions = defaultdict(list)
1
6f1378738798 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_signatures commit 6133bb114c76a795fa12a4a11edb1a8b80fd104d
artbio
parents:
diff changeset
50 for chrom in self.chromosomes:
6f1378738798 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_signatures commit 6133bb114c76a795fa12a4a11edb1a8b80fd104d
artbio
parents:
diff changeset
51 for read in bam_object.fetch(chrom):
3
4d9682bd3a6b planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_signatures commit 96ed5824190aff281cc3aa47dc60fc66aac41db3
artbio
parents: 1
diff changeset
52 if not read.is_reverse:
4d9682bd3a6b planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_signatures commit 96ed5824190aff281cc3aa47dc60fc66aac41db3
artbio
parents: 1
diff changeset
53 all_query_positions[chrom].append(
4
20d28cfdeefe planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_signatures commit cfdc08418887bfe4a35588cd78d0a2b6ffa6e19e
artbio
parents: 3
diff changeset
54 read.reference_start)
1
6f1378738798 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_signatures commit 6133bb114c76a795fa12a4a11edb1a8b80fd104d
artbio
parents:
diff changeset
55 else:
3
4d9682bd3a6b planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_signatures commit 96ed5824190aff281cc3aa47dc60fc66aac41db3
artbio
parents: 1
diff changeset
56 all_query_positions[chrom].append(
4
20d28cfdeefe planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_signatures commit cfdc08418887bfe4a35588cd78d0a2b6ffa6e19e
artbio
parents: 3
diff changeset
57 read.reference_end)
3
4d9682bd3a6b planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_signatures commit 96ed5824190aff281cc3aa47dc60fc66aac41db3
artbio
parents: 1
diff changeset
58 all_query_positions[chrom] = sorted(
4d9682bd3a6b planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_signatures commit 96ed5824190aff281cc3aa47dc60fc66aac41db3
artbio
parents: 1
diff changeset
59 list(set(all_query_positions[chrom])))
4d9682bd3a6b planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_signatures commit 96ed5824190aff281cc3aa47dc60fc66aac41db3
artbio
parents: 1
diff changeset
60 return all_query_positions
1
6f1378738798 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_signatures commit 6133bb114c76a795fa12a4a11edb1a8b80fd104d
artbio
parents:
diff changeset
61
3
4d9682bd3a6b planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_signatures commit 96ed5824190aff281cc3aa47dc60fc66aac41db3
artbio
parents: 1
diff changeset
62 def direct_pairing(self, minquery, maxquery, mintarget, maxtarget,
4d9682bd3a6b planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_signatures commit 96ed5824190aff281cc3aa47dc60fc66aac41db3
artbio
parents: 1
diff changeset
63 file, overlap=10):
4d9682bd3a6b planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_signatures commit 96ed5824190aff281cc3aa47dc60fc66aac41db3
artbio
parents: 1
diff changeset
64 F = open(file, 'w')
1
6f1378738798 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_signatures commit 6133bb114c76a795fa12a4a11edb1a8b80fd104d
artbio
parents:
diff changeset
65 query_range = range(minquery, maxquery + 1)
6f1378738798 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_signatures commit 6133bb114c76a795fa12a4a11edb1a8b80fd104d
artbio
parents:
diff changeset
66 target_range = range(mintarget, maxtarget + 1)
3
4d9682bd3a6b planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_signatures commit 96ed5824190aff281cc3aa47dc60fc66aac41db3
artbio
parents: 1
diff changeset
67 stringresult = []
4d9682bd3a6b planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_signatures commit 96ed5824190aff281cc3aa47dc60fc66aac41db3
artbio
parents: 1
diff changeset
68 for chrom in sorted(self.chromosomes):
4d9682bd3a6b planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_signatures commit 96ed5824190aff281cc3aa47dc60fc66aac41db3
artbio
parents: 1
diff changeset
69 for pos in (self.all_query_positions[chrom]):
4d9682bd3a6b planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_signatures commit 96ed5824190aff281cc3aa47dc60fc66aac41db3
artbio
parents: 1
diff changeset
70 iterreads_1 = self.bam_object.fetch(chrom,
4d9682bd3a6b planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_signatures commit 96ed5824190aff281cc3aa47dc60fc66aac41db3
artbio
parents: 1
diff changeset
71 start=pos, end=pos+overlap-1)
4d9682bd3a6b planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_signatures commit 96ed5824190aff281cc3aa47dc60fc66aac41db3
artbio
parents: 1
diff changeset
72 iterreads_2 = self.bam_object.fetch(chrom,
4d9682bd3a6b planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_signatures commit 96ed5824190aff281cc3aa47dc60fc66aac41db3
artbio
parents: 1
diff changeset
73 start=pos, end=pos+overlap-1)
4d9682bd3a6b planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_signatures commit 96ed5824190aff281cc3aa47dc60fc66aac41db3
artbio
parents: 1
diff changeset
74 iterreads_3 = self.bam_object.fetch(chrom,
4d9682bd3a6b planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_signatures commit 96ed5824190aff281cc3aa47dc60fc66aac41db3
artbio
parents: 1
diff changeset
75 start=pos, end=pos+overlap-1)
4d9682bd3a6b planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_signatures commit 96ed5824190aff281cc3aa47dc60fc66aac41db3
artbio
parents: 1
diff changeset
76 iterreads_4 = self.bam_object.fetch(chrom,
4d9682bd3a6b planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_signatures commit 96ed5824190aff281cc3aa47dc60fc66aac41db3
artbio
parents: 1
diff changeset
77 start=pos, end=pos+overlap-1)
4d9682bd3a6b planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_signatures commit 96ed5824190aff281cc3aa47dc60fc66aac41db3
artbio
parents: 1
diff changeset
78 # 1
4d9682bd3a6b planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_signatures commit 96ed5824190aff281cc3aa47dc60fc66aac41db3
artbio
parents: 1
diff changeset
79 for queryread in iterreads_1:
4
20d28cfdeefe planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_signatures commit cfdc08418887bfe4a35588cd78d0a2b6ffa6e19e
artbio
parents: 3
diff changeset
80 if queryread.reference_start == pos and \
3
4d9682bd3a6b planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_signatures commit 96ed5824190aff281cc3aa47dc60fc66aac41db3
artbio
parents: 1
diff changeset
81 queryread.query_alignment_length in query_range \
4d9682bd3a6b planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_signatures commit 96ed5824190aff281cc3aa47dc60fc66aac41db3
artbio
parents: 1
diff changeset
82 and not queryread.is_reverse:
4d9682bd3a6b planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_signatures commit 96ed5824190aff281cc3aa47dc60fc66aac41db3
artbio
parents: 1
diff changeset
83 for targetread in iterreads_2:
4d9682bd3a6b planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_signatures commit 96ed5824190aff281cc3aa47dc60fc66aac41db3
artbio
parents: 1
diff changeset
84 if (targetread.
4
20d28cfdeefe planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_signatures commit cfdc08418887bfe4a35588cd78d0a2b6ffa6e19e
artbio
parents: 3
diff changeset
85 get_reference_positions()[-1]
3
4d9682bd3a6b planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_signatures commit 96ed5824190aff281cc3aa47dc60fc66aac41db3
artbio
parents: 1
diff changeset
86 == queryread.get_reference_positions(
4
20d28cfdeefe planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_signatures commit cfdc08418887bfe4a35588cd78d0a2b6ffa6e19e
artbio
parents: 3
diff changeset
87 )[overlap-1] and
3
4d9682bd3a6b planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_signatures commit 96ed5824190aff281cc3aa47dc60fc66aac41db3
artbio
parents: 1
diff changeset
88 targetread.query_alignment_length in
4d9682bd3a6b planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_signatures commit 96ed5824190aff281cc3aa47dc60fc66aac41db3
artbio
parents: 1
diff changeset
89 target_range and targetread.is_reverse):
4d9682bd3a6b planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_signatures commit 96ed5824190aff281cc3aa47dc60fc66aac41db3
artbio
parents: 1
diff changeset
90 targetreadseq = self.revcomp(
4d9682bd3a6b planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_signatures commit 96ed5824190aff281cc3aa47dc60fc66aac41db3
artbio
parents: 1
diff changeset
91 targetread.query_sequence)
4d9682bd3a6b planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_signatures commit 96ed5824190aff281cc3aa47dc60fc66aac41db3
artbio
parents: 1
diff changeset
92 stringresult.append(
4d9682bd3a6b planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_signatures commit 96ed5824190aff281cc3aa47dc60fc66aac41db3
artbio
parents: 1
diff changeset
93 '>%s|%s|%s|%s|n=%s\n%s\n' %
4
20d28cfdeefe planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_signatures commit cfdc08418887bfe4a35588cd78d0a2b6ffa6e19e
artbio
parents: 3
diff changeset
94 (chrom, queryread.reference_start+1,
3
4d9682bd3a6b planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_signatures commit 96ed5824190aff281cc3aa47dc60fc66aac41db3
artbio
parents: 1
diff changeset
95 'F', queryread.query_alignment_length,
4d9682bd3a6b planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_signatures commit 96ed5824190aff281cc3aa47dc60fc66aac41db3
artbio
parents: 1
diff changeset
96 self.readdic[queryread.query_sequence],
4d9682bd3a6b planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_signatures commit 96ed5824190aff281cc3aa47dc60fc66aac41db3
artbio
parents: 1
diff changeset
97 queryread.query_sequence))
4d9682bd3a6b planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_signatures commit 96ed5824190aff281cc3aa47dc60fc66aac41db3
artbio
parents: 1
diff changeset
98 stringresult.append(
4d9682bd3a6b planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_signatures commit 96ed5824190aff281cc3aa47dc60fc66aac41db3
artbio
parents: 1
diff changeset
99 '>%s|%s|%s|%s|n=%s\n%s\n' %
4
20d28cfdeefe planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_signatures commit cfdc08418887bfe4a35588cd78d0a2b6ffa6e19e
artbio
parents: 3
diff changeset
100 (chrom, targetread.reference_start+1,
3
4d9682bd3a6b planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_signatures commit 96ed5824190aff281cc3aa47dc60fc66aac41db3
artbio
parents: 1
diff changeset
101 'R', targetread.query_alignment_length,
4d9682bd3a6b planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_signatures commit 96ed5824190aff281cc3aa47dc60fc66aac41db3
artbio
parents: 1
diff changeset
102 self.readdic[targetread.query_sequence],
4d9682bd3a6b planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_signatures commit 96ed5824190aff281cc3aa47dc60fc66aac41db3
artbio
parents: 1
diff changeset
103 targetreadseq))
4d9682bd3a6b planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_signatures commit 96ed5824190aff281cc3aa47dc60fc66aac41db3
artbio
parents: 1
diff changeset
104 # 2
4d9682bd3a6b planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_signatures commit 96ed5824190aff281cc3aa47dc60fc66aac41db3
artbio
parents: 1
diff changeset
105 for queryread in iterreads_3:
4
20d28cfdeefe planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_signatures commit cfdc08418887bfe4a35588cd78d0a2b6ffa6e19e
artbio
parents: 3
diff changeset
106 if queryread.reference_end-1 == pos+overlap-1 and \
3
4d9682bd3a6b planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_signatures commit 96ed5824190aff281cc3aa47dc60fc66aac41db3
artbio
parents: 1
diff changeset
107 queryread.query_alignment_length in query_range \
4d9682bd3a6b planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_signatures commit 96ed5824190aff281cc3aa47dc60fc66aac41db3
artbio
parents: 1
diff changeset
108 and queryread.is_reverse:
4d9682bd3a6b planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_signatures commit 96ed5824190aff281cc3aa47dc60fc66aac41db3
artbio
parents: 1
diff changeset
109 for targetread in iterreads_4:
4d9682bd3a6b planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_signatures commit 96ed5824190aff281cc3aa47dc60fc66aac41db3
artbio
parents: 1
diff changeset
110 if (targetread.
4
20d28cfdeefe planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_signatures commit cfdc08418887bfe4a35588cd78d0a2b6ffa6e19e
artbio
parents: 3
diff changeset
111 reference_start
3
4d9682bd3a6b planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_signatures commit 96ed5824190aff281cc3aa47dc60fc66aac41db3
artbio
parents: 1
diff changeset
112 == pos and targetread.query_alignment_length
4d9682bd3a6b planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_signatures commit 96ed5824190aff281cc3aa47dc60fc66aac41db3
artbio
parents: 1
diff changeset
113 in target_range and not
4d9682bd3a6b planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_signatures commit 96ed5824190aff281cc3aa47dc60fc66aac41db3
artbio
parents: 1
diff changeset
114 targetread.is_reverse):
4d9682bd3a6b planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_signatures commit 96ed5824190aff281cc3aa47dc60fc66aac41db3
artbio
parents: 1
diff changeset
115 queryreadseq = self.revcomp(
4d9682bd3a6b planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_signatures commit 96ed5824190aff281cc3aa47dc60fc66aac41db3
artbio
parents: 1
diff changeset
116 queryread.query_sequence)
4d9682bd3a6b planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_signatures commit 96ed5824190aff281cc3aa47dc60fc66aac41db3
artbio
parents: 1
diff changeset
117 targetreadseq = targetread.query_sequence
4d9682bd3a6b planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_signatures commit 96ed5824190aff281cc3aa47dc60fc66aac41db3
artbio
parents: 1
diff changeset
118 stringresult.append(
4d9682bd3a6b planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_signatures commit 96ed5824190aff281cc3aa47dc60fc66aac41db3
artbio
parents: 1
diff changeset
119 '>%s|%s|%s|%s|n=%s\n%s\n' %
4
20d28cfdeefe planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_signatures commit cfdc08418887bfe4a35588cd78d0a2b6ffa6e19e
artbio
parents: 3
diff changeset
120 (chrom, queryread.reference_start+1, 'R',
3
4d9682bd3a6b planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_signatures commit 96ed5824190aff281cc3aa47dc60fc66aac41db3
artbio
parents: 1
diff changeset
121 queryread.query_alignment_length,
4d9682bd3a6b planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_signatures commit 96ed5824190aff281cc3aa47dc60fc66aac41db3
artbio
parents: 1
diff changeset
122 self.readdic[queryread.query_sequence],
4d9682bd3a6b planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_signatures commit 96ed5824190aff281cc3aa47dc60fc66aac41db3
artbio
parents: 1
diff changeset
123 queryreadseq))
4d9682bd3a6b planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_signatures commit 96ed5824190aff281cc3aa47dc60fc66aac41db3
artbio
parents: 1
diff changeset
124 stringresult.append(
4d9682bd3a6b planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_signatures commit 96ed5824190aff281cc3aa47dc60fc66aac41db3
artbio
parents: 1
diff changeset
125 '>%s|%s|%s|%s|n=%s\n%s\n' %
4
20d28cfdeefe planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_signatures commit cfdc08418887bfe4a35588cd78d0a2b6ffa6e19e
artbio
parents: 3
diff changeset
126 (chrom, targetread.reference_start+1,
3
4d9682bd3a6b planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_signatures commit 96ed5824190aff281cc3aa47dc60fc66aac41db3
artbio
parents: 1
diff changeset
127 'F', targetread.query_alignment_length,
4d9682bd3a6b planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_signatures commit 96ed5824190aff281cc3aa47dc60fc66aac41db3
artbio
parents: 1
diff changeset
128 self.readdic[targetread.query_sequence],
4d9682bd3a6b planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_signatures commit 96ed5824190aff281cc3aa47dc60fc66aac41db3
artbio
parents: 1
diff changeset
129 targetreadseq))
4d9682bd3a6b planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_signatures commit 96ed5824190aff281cc3aa47dc60fc66aac41db3
artbio
parents: 1
diff changeset
130 stringresult = sorted(set(stringresult),
4d9682bd3a6b planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_signatures commit 96ed5824190aff281cc3aa47dc60fc66aac41db3
artbio
parents: 1
diff changeset
131 key=lambda x: stringresult.index(x))
4d9682bd3a6b planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_signatures commit 96ed5824190aff281cc3aa47dc60fc66aac41db3
artbio
parents: 1
diff changeset
132 F.write(''.join(stringresult))
1
6f1378738798 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_signatures commit 6133bb114c76a795fa12a4a11edb1a8b80fd104d
artbio
parents:
diff changeset
133
6f1378738798 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_signatures commit 6133bb114c76a795fa12a4a11edb1a8b80fd104d
artbio
parents:
diff changeset
134 def revcomp(self, sequence):
6f1378738798 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_signatures commit 6133bb114c76a795fa12a4a11edb1a8b80fd104d
artbio
parents:
diff changeset
135 antidict = {"A": "T", "T": "A", "G": "C", "C": "G", "N": "N"}
6f1378738798 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_signatures commit 6133bb114c76a795fa12a4a11edb1a8b80fd104d
artbio
parents:
diff changeset
136 revseq = sequence[::-1]
6f1378738798 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_signatures commit 6133bb114c76a795fa12a4a11edb1a8b80fd104d
artbio
parents:
diff changeset
137 return "".join([antidict[i] for i in revseq])
6f1378738798 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_signatures commit 6133bb114c76a795fa12a4a11edb1a8b80fd104d
artbio
parents:
diff changeset
138
6f1378738798 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_signatures commit 6133bb114c76a795fa12a4a11edb1a8b80fd104d
artbio
parents:
diff changeset
139
6f1378738798 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_signatures commit 6133bb114c76a795fa12a4a11edb1a8b80fd104d
artbio
parents:
diff changeset
140 def main(input, minquery, maxquery, mintarget, maxtarget, output, overlap=10):
6f1378738798 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_signatures commit 6133bb114c76a795fa12a4a11edb1a8b80fd104d
artbio
parents:
diff changeset
141 mapobj = Map(input)
3
4d9682bd3a6b planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_signatures commit 96ed5824190aff281cc3aa47dc60fc66aac41db3
artbio
parents: 1
diff changeset
142 mapobj.direct_pairing(minquery, maxquery, mintarget, maxtarget,
4d9682bd3a6b planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_signatures commit 96ed5824190aff281cc3aa47dc60fc66aac41db3
artbio
parents: 1
diff changeset
143 output, overlap)
1
6f1378738798 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_signatures commit 6133bb114c76a795fa12a4a11edb1a8b80fd104d
artbio
parents:
diff changeset
144
6f1378738798 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_signatures commit 6133bb114c76a795fa12a4a11edb1a8b80fd104d
artbio
parents:
diff changeset
145
6f1378738798 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_signatures commit 6133bb114c76a795fa12a4a11edb1a8b80fd104d
artbio
parents:
diff changeset
146 if __name__ == "__main__":
6f1378738798 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_signatures commit 6133bb114c76a795fa12a4a11edb1a8b80fd104d
artbio
parents:
diff changeset
147 args = Parser()
6f1378738798 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_signatures commit 6133bb114c76a795fa12a4a11edb1a8b80fd104d
artbio
parents:
diff changeset
148 main(args.input, args.minquery, args.maxquery, args.mintarget,
3
4d9682bd3a6b planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_signatures commit 96ed5824190aff281cc3aa47dc60fc66aac41db3
artbio
parents: 1
diff changeset
149 args.maxtarget, args.output, args.overlap)