Mercurial > repos > artbio > small_rna_signatures

annotate overlapping_reads.py @ 1:6f1378738798 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 6133bb114c76a795fa12a4a11edb1a8b80fd104d
author artbio
date Tue, 29 Aug 2017 20:02:15 -0400
parents
children 4d9682bd3a6b
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))
6f1378738798 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_signatures commit 6133bb114c76a795fa12a4a11edb1a8b80fd104d
artbio
parents:
diff changeset
39 self.map_dict = self.create_map(self.bam_object)
6f1378738798 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_signatures commit 6133bb114c76a795fa12a4a11edb1a8b80fd104d
artbio
parents:
diff changeset
40
6f1378738798 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_signatures commit 6133bb114c76a795fa12a4a11edb1a8b80fd104d
artbio
parents:
diff changeset
41 def create_map(self, bam_object):
6f1378738798 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_signatures commit 6133bb114c76a795fa12a4a11edb1a8b80fd104d
artbio
parents:
diff changeset
42 '''
6f1378738798 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_signatures commit 6133bb114c76a795fa12a4a11edb1a8b80fd104d
artbio
parents:
diff changeset
43 Returns a map_dictionary {(chromosome,read_position,polarity):
6f1378738798 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_signatures commit 6133bb114c76a795fa12a4a11edb1a8b80fd104d
artbio
parents:
diff changeset
44 [read_length, ...]}
6f1378738798 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_signatures commit 6133bb114c76a795fa12a4a11edb1a8b80fd104d
artbio
parents:
diff changeset
45 '''
6f1378738798 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_signatures commit 6133bb114c76a795fa12a4a11edb1a8b80fd104d
artbio
parents:
diff changeset
46 map_dictionary = defaultdict(list)
6f1378738798 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_signatures commit 6133bb114c76a795fa12a4a11edb1a8b80fd104d
artbio
parents:
diff changeset
47 # get empty value for start and end of each chromosome
6f1378738798 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_signatures commit 6133bb114c76a795fa12a4a11edb1a8b80fd104d
artbio
parents:
diff changeset
48 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
49 map_dictionary[(chrom, 1, 'F')] = []
6f1378738798 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_signatures commit 6133bb114c76a795fa12a4a11edb1a8b80fd104d
artbio
parents:
diff changeset
50 map_dictionary[(chrom, self.chromosomes[chrom], 'F')] = []
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 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
52 for read in bam_object.fetch(chrom):
6f1378738798 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_signatures commit 6133bb114c76a795fa12a4a11edb1a8b80fd104d
artbio
parents:
diff changeset
53 positions = read.positions # a list of covered positions
6f1378738798 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_signatures commit 6133bb114c76a795fa12a4a11edb1a8b80fd104d
artbio
parents:
diff changeset
54 if read.is_reverse:
6f1378738798 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_signatures commit 6133bb114c76a795fa12a4a11edb1a8b80fd104d
artbio
parents:
diff changeset
55 map_dictionary[(chrom, positions[-1]+1,
6f1378738798 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_signatures commit 6133bb114c76a795fa12a4a11edb1a8b80fd104d
artbio
parents:
diff changeset
56 'R')].append(read.query_alignment_length)
6f1378738798 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_signatures commit 6133bb114c76a795fa12a4a11edb1a8b80fd104d
artbio
parents:
diff changeset
57 else:
6f1378738798 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_signatures commit 6133bb114c76a795fa12a4a11edb1a8b80fd104d
artbio
parents:
diff changeset
58 map_dictionary[(chrom, positions[0]+1,
6f1378738798 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_signatures commit 6133bb114c76a795fa12a4a11edb1a8b80fd104d
artbio
parents:
diff changeset
59 'F')].append(read.query_alignment_length)
6f1378738798 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_signatures commit 6133bb114c76a795fa12a4a11edb1a8b80fd104d
artbio
parents:
diff changeset
60 return map_dictionary
6f1378738798 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_signatures commit 6133bb114c76a795fa12a4a11edb1a8b80fd104d
artbio
parents:
diff changeset
61
6f1378738798 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_signatures commit 6133bb114c76a795fa12a4a11edb1a8b80fd104d
artbio
parents:
diff changeset
62 def signature_tables(self, minquery, maxquery, mintarget, maxtarget):
6f1378738798 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_signatures commit 6133bb114c76a795fa12a4a11edb1a8b80fd104d
artbio
parents:
diff changeset
63 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
64 target_range = range(mintarget, maxtarget + 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_table = defaultdict(dict)
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_table = defaultdict(dict)
6f1378738798 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_signatures commit 6133bb114c76a795fa12a4a11edb1a8b80fd104d
artbio
parents:
diff changeset
67 for key in self.map_dict:
6f1378738798 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_signatures commit 6133bb114c76a795fa12a4a11edb1a8b80fd104d
artbio
parents:
diff changeset
68 for size in self.map_dict[key]:
6f1378738798 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_signatures commit 6133bb114c76a795fa12a4a11edb1a8b80fd104d
artbio
parents:
diff changeset
69 if size in query_range or size in target_range:
6f1378738798 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_signatures commit 6133bb114c76a795fa12a4a11edb1a8b80fd104d
artbio
parents:
diff changeset
70 if key[2] == 'F':
6f1378738798 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_signatures commit 6133bb114c76a795fa12a4a11edb1a8b80fd104d
artbio
parents:
diff changeset
71 coordinate = key[1]
6f1378738798 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_signatures commit 6133bb114c76a795fa12a4a11edb1a8b80fd104d
artbio
parents:
diff changeset
72 else:
6f1378738798 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_signatures commit 6133bb114c76a795fa12a4a11edb1a8b80fd104d
artbio
parents:
diff changeset
73 coordinate = -key[1]
6f1378738798 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_signatures commit 6133bb114c76a795fa12a4a11edb1a8b80fd104d
artbio
parents:
diff changeset
74 if size in query_range:
6f1378738798 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_signatures commit 6133bb114c76a795fa12a4a11edb1a8b80fd104d
artbio
parents:
diff changeset
75 Query_table[key[0]][coordinate] = Query_table[key[0]].get(
6f1378738798 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_signatures commit 6133bb114c76a795fa12a4a11edb1a8b80fd104d
artbio
parents:
diff changeset
76 coordinate, 0) + 1
6f1378738798 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_signatures commit 6133bb114c76a795fa12a4a11edb1a8b80fd104d
artbio
parents:
diff changeset
77 if size in target_range:
6f1378738798 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_signatures commit 6133bb114c76a795fa12a4a11edb1a8b80fd104d
artbio
parents:
diff changeset
78 Target_table[key[0]][coordinate] = \
6f1378738798 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_signatures commit 6133bb114c76a795fa12a4a11edb1a8b80fd104d
artbio
parents:
diff changeset
79 Target_table[key[0]].get(coordinate, 0) + 1
6f1378738798 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_signatures commit 6133bb114c76a795fa12a4a11edb1a8b80fd104d
artbio
parents:
diff changeset
80 return Query_table, Target_table
6f1378738798 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_signatures commit 6133bb114c76a795fa12a4a11edb1a8b80fd104d
artbio
parents:
diff changeset
81
6f1378738798 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_signatures commit 6133bb114c76a795fa12a4a11edb1a8b80fd104d
artbio
parents:
diff changeset
82 def search_overlaps(self, minquery, maxquery, mintarget, maxtarget,
6f1378738798 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_signatures commit 6133bb114c76a795fa12a4a11edb1a8b80fd104d
artbio
parents:
diff changeset
83 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
84 Query_table, Target_table = self.signature_tables(minquery, maxquery,
6f1378738798 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_signatures commit 6133bb114c76a795fa12a4a11edb1a8b80fd104d
artbio
parents:
diff changeset
85 mintarget, maxtarget)
6f1378738798 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_signatures commit 6133bb114c76a795fa12a4a11edb1a8b80fd104d
artbio
parents:
diff changeset
86 overlap_groups = defaultdict(list)
6f1378738798 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_signatures commit 6133bb114c76a795fa12a4a11edb1a8b80fd104d
artbio
parents:
diff changeset
87 for chrom in Query_table:
6f1378738798 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_signatures commit 6133bb114c76a795fa12a4a11edb1a8b80fd104d
artbio
parents:
diff changeset
88 for coord in Query_table[chrom]:
6f1378738798 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_signatures commit 6133bb114c76a795fa12a4a11edb1a8b80fd104d
artbio
parents:
diff changeset
89 if Target_table[chrom].get(-coord - overlap + 1, 0):
6f1378738798 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_signatures commit 6133bb114c76a795fa12a4a11edb1a8b80fd104d
artbio
parents:
diff changeset
90 overlap_groups[chrom].append(coord)
6f1378738798 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_signatures commit 6133bb114c76a795fa12a4a11edb1a8b80fd104d
artbio
parents:
diff changeset
91 return overlap_groups
6f1378738798 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_signatures commit 6133bb114c76a795fa12a4a11edb1a8b80fd104d
artbio
parents:
diff changeset
92
6f1378738798 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_signatures commit 6133bb114c76a795fa12a4a11edb1a8b80fd104d
artbio
parents:
diff changeset
93 def feed_overlaps(self, overlap_groups, minquery, 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
94 F = open(output, 'w')
6f1378738798 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_signatures commit 6133bb114c76a795fa12a4a11edb1a8b80fd104d
artbio
parents:
diff changeset
95 for chrom in sorted(overlap_groups):
6f1378738798 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_signatures commit 6133bb114c76a795fa12a4a11edb1a8b80fd104d
artbio
parents:
diff changeset
96 for pos in sorted(overlap_groups[chrom]):
6f1378738798 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_signatures commit 6133bb114c76a795fa12a4a11edb1a8b80fd104d
artbio
parents:
diff changeset
97 if pos > 0: # read are forward
6f1378738798 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_signatures commit 6133bb114c76a795fa12a4a11edb1a8b80fd104d
artbio
parents:
diff changeset
98 reads = self.bam_object.fetch(chrom, start=pos-1,
6f1378738798 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_signatures commit 6133bb114c76a795fa12a4a11edb1a8b80fd104d
artbio
parents:
diff changeset
99 end=pos-1+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
100 for read in reads:
6f1378738798 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_signatures commit 6133bb114c76a795fa12a4a11edb1a8b80fd104d
artbio
parents:
diff changeset
101 positions = read.positions
6f1378738798 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_signatures commit 6133bb114c76a795fa12a4a11edb1a8b80fd104d
artbio
parents:
diff changeset
102 if pos-1 == positions[0] and \
6f1378738798 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_signatures commit 6133bb114c76a795fa12a4a11edb1a8b80fd104d
artbio
parents:
diff changeset
103 read.query_alignment_length >= minquery:
6f1378738798 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_signatures commit 6133bb114c76a795fa12a4a11edb1a8b80fd104d
artbio
parents:
diff changeset
104 F.write('>%s|%s|%s|%s\n%s\n' % (
6f1378738798 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_signatures commit 6133bb114c76a795fa12a4a11edb1a8b80fd104d
artbio
parents:
diff changeset
105 chrom, pos, 'F',
6f1378738798 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_signatures commit 6133bb114c76a795fa12a4a11edb1a8b80fd104d
artbio
parents:
diff changeset
106 read.query_alignment_length,
6f1378738798 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_signatures commit 6133bb114c76a795fa12a4a11edb1a8b80fd104d
artbio
parents:
diff changeset
107 read.query_sequence))
6f1378738798 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_signatures commit 6133bb114c76a795fa12a4a11edb1a8b80fd104d
artbio
parents:
diff changeset
108 else: # reads are reverse
6f1378738798 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_signatures commit 6133bb114c76a795fa12a4a11edb1a8b80fd104d
artbio
parents:
diff changeset
109 reads = self.bam_object.fetch(chrom,
6f1378738798 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_signatures commit 6133bb114c76a795fa12a4a11edb1a8b80fd104d
artbio
parents:
diff changeset
110 start=-pos-1-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
111 end=-pos-1)
6f1378738798 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_signatures commit 6133bb114c76a795fa12a4a11edb1a8b80fd104d
artbio
parents:
diff changeset
112 for read in reads:
6f1378738798 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_signatures commit 6133bb114c76a795fa12a4a11edb1a8b80fd104d
artbio
parents:
diff changeset
113 positions = read.positions
6f1378738798 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_signatures commit 6133bb114c76a795fa12a4a11edb1a8b80fd104d
artbio
parents:
diff changeset
114 if -pos-1 == positions[-1] and \
6f1378738798 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_signatures commit 6133bb114c76a795fa12a4a11edb1a8b80fd104d
artbio
parents:
diff changeset
115 read.query_alignment_length >= minquery:
6f1378738798 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_signatures commit 6133bb114c76a795fa12a4a11edb1a8b80fd104d
artbio
parents:
diff changeset
116 readseq = self.revcomp(read.query_sequence)
6f1378738798 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_signatures commit 6133bb114c76a795fa12a4a11edb1a8b80fd104d
artbio
parents:
diff changeset
117 readsize = read.query_alignment_length
6f1378738798 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_signatures commit 6133bb114c76a795fa12a4a11edb1a8b80fd104d
artbio
parents:
diff changeset
118 F.write('>%s|%s|%s|%s\n%s\n' % (chrom,
6f1378738798 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_signatures commit 6133bb114c76a795fa12a4a11edb1a8b80fd104d
artbio
parents:
diff changeset
119 positions[0] + 1,
6f1378738798 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_signatures commit 6133bb114c76a795fa12a4a11edb1a8b80fd104d
artbio
parents:
diff changeset
120 'R', readsize, readseq))
6f1378738798 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_signatures commit 6133bb114c76a795fa12a4a11edb1a8b80fd104d
artbio
parents:
diff changeset
121 F.close()
6f1378738798 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_signatures commit 6133bb114c76a795fa12a4a11edb1a8b80fd104d
artbio
parents:
diff changeset
122 return
6f1378738798 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_signatures commit 6133bb114c76a795fa12a4a11edb1a8b80fd104d
artbio
parents:
diff changeset
123
6f1378738798 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_signatures commit 6133bb114c76a795fa12a4a11edb1a8b80fd104d
artbio
parents:
diff changeset
124 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
125 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
126 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
127 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
128
6f1378738798 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_signatures commit 6133bb114c76a795fa12a4a11edb1a8b80fd104d
artbio
parents:
diff changeset
129
6f1378738798 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_signatures commit 6133bb114c76a795fa12a4a11edb1a8b80fd104d
artbio
parents:
diff changeset
130 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
131 mapobj = Map(input)
6f1378738798 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_signatures commit 6133bb114c76a795fa12a4a11edb1a8b80fd104d
artbio
parents:
diff changeset
132 mapobj.feed_overlaps(mapobj.search_overlaps(minquery, maxquery,
6f1378738798 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_signatures commit 6133bb114c76a795fa12a4a11edb1a8b80fd104d
artbio
parents:
diff changeset
133 mintarget, maxtarget,
6f1378738798 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_signatures commit 6133bb114c76a795fa12a4a11edb1a8b80fd104d
artbio
parents:
diff changeset
134 overlap), minquery, output)
6f1378738798 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_signatures commit 6133bb114c76a795fa12a4a11edb1a8b80fd104d
artbio
parents:
diff changeset
135
6f1378738798 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_signatures commit 6133bb114c76a795fa12a4a11edb1a8b80fd104d
artbio
parents:
diff changeset
136
6f1378738798 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_signatures commit 6133bb114c76a795fa12a4a11edb1a8b80fd104d
artbio
parents:
diff changeset
137 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
138 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
139 main(args.input, args.minquery, args.maxquery, args.mintarget,
6f1378738798 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_signatures commit 6133bb114c76a795fa12a4a11edb1a8b80fd104d
artbio
parents:
diff changeset
140 args.maxtarget, args.output)