Mercurial > repos > artbio > small_rna_signatures

annotate signature.py @ 0:a35e6f9c1d34 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 6719543c5017d581ae012b864d7c9088f0767fc8
author artbio
date Mon, 28 Aug 2017 09:29:47 -0400
parents
children 07771982ef9b
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
rev   line source
0
a35e6f9c1d34 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_signatures commit 6719543c5017d581ae012b864d7c9088f0767fc8
artbio
parents:
diff changeset
1 import argparse
a35e6f9c1d34 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_signatures commit 6719543c5017d581ae012b864d7c9088f0767fc8
artbio
parents:
diff changeset
2 from collections import defaultdict
a35e6f9c1d34 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_signatures commit 6719543c5017d581ae012b864d7c9088f0767fc8
artbio
parents:
diff changeset
3
a35e6f9c1d34 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_signatures commit 6719543c5017d581ae012b864d7c9088f0767fc8
artbio
parents:
diff changeset
4 import numpy
a35e6f9c1d34 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_signatures commit 6719543c5017d581ae012b864d7c9088f0767fc8
artbio
parents:
diff changeset
5
a35e6f9c1d34 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_signatures commit 6719543c5017d581ae012b864d7c9088f0767fc8
artbio
parents:
diff changeset
6 import pysam
a35e6f9c1d34 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_signatures commit 6719543c5017d581ae012b864d7c9088f0767fc8
artbio
parents:
diff changeset
7
a35e6f9c1d34 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_signatures commit 6719543c5017d581ae012b864d7c9088f0767fc8
artbio
parents:
diff changeset
8
a35e6f9c1d34 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_signatures commit 6719543c5017d581ae012b864d7c9088f0767fc8
artbio
parents:
diff changeset
9 def Parser():
a35e6f9c1d34 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_signatures commit 6719543c5017d581ae012b864d7c9088f0767fc8
artbio
parents:
diff changeset
10 the_parser = argparse.ArgumentParser()
a35e6f9c1d34 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_signatures commit 6719543c5017d581ae012b864d7c9088f0767fc8
artbio
parents:
diff changeset
11 the_parser.add_argument(
a35e6f9c1d34 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_signatures commit 6719543c5017d581ae012b864d7c9088f0767fc8
artbio
parents:
diff changeset
12 '--input', action="store", type=str, help="bam alignment file")
a35e6f9c1d34 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_signatures commit 6719543c5017d581ae012b864d7c9088f0767fc8
artbio
parents:
diff changeset
13 the_parser.add_argument(
a35e6f9c1d34 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_signatures commit 6719543c5017d581ae012b864d7c9088f0767fc8
artbio
parents:
diff changeset
14 '--minquery', type=int,
a35e6f9c1d34 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_signatures commit 6719543c5017d581ae012b864d7c9088f0767fc8
artbio
parents:
diff changeset
15 help="Minimum readsize of query reads (nt) - must be an integer")
a35e6f9c1d34 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_signatures commit 6719543c5017d581ae012b864d7c9088f0767fc8
artbio
parents:
diff changeset
16 the_parser.add_argument(
a35e6f9c1d34 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_signatures commit 6719543c5017d581ae012b864d7c9088f0767fc8
artbio
parents:
diff changeset
17 '--maxquery', type=int,
a35e6f9c1d34 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_signatures commit 6719543c5017d581ae012b864d7c9088f0767fc8
artbio
parents:
diff changeset
18 help="Maximum readsize of query reads (nt) - must be an integer")
a35e6f9c1d34 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_signatures commit 6719543c5017d581ae012b864d7c9088f0767fc8
artbio
parents:
diff changeset
19 the_parser.add_argument(
a35e6f9c1d34 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_signatures commit 6719543c5017d581ae012b864d7c9088f0767fc8
artbio
parents:
diff changeset
20 '--mintarget', type=int,
a35e6f9c1d34 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_signatures commit 6719543c5017d581ae012b864d7c9088f0767fc8
artbio
parents:
diff changeset
21 help="Minimum readsize of target reads (nt) - must be an integer")
a35e6f9c1d34 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_signatures commit 6719543c5017d581ae012b864d7c9088f0767fc8
artbio
parents:
diff changeset
22 the_parser.add_argument(
a35e6f9c1d34 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_signatures commit 6719543c5017d581ae012b864d7c9088f0767fc8
artbio
parents:
diff changeset
23 '--maxtarget', type=int,
a35e6f9c1d34 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_signatures commit 6719543c5017d581ae012b864d7c9088f0767fc8
artbio
parents:
diff changeset
24 help="Maximum readsize of target reads (nt) - must be an integer")
a35e6f9c1d34 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_signatures commit 6719543c5017d581ae012b864d7c9088f0767fc8
artbio
parents:
diff changeset
25 the_parser.add_argument(
a35e6f9c1d34 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_signatures commit 6719543c5017d581ae012b864d7c9088f0767fc8
artbio
parents:
diff changeset
26 '--minscope', type=int,
a35e6f9c1d34 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_signatures commit 6719543c5017d581ae012b864d7c9088f0767fc8
artbio
parents:
diff changeset
27 help="Minimum overlap analyzed (nt) - must be an integer")
a35e6f9c1d34 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_signatures commit 6719543c5017d581ae012b864d7c9088f0767fc8
artbio
parents:
diff changeset
28 the_parser.add_argument(
a35e6f9c1d34 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_signatures commit 6719543c5017d581ae012b864d7c9088f0767fc8
artbio
parents:
diff changeset
29 '--maxscope', type=int,
a35e6f9c1d34 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_signatures commit 6719543c5017d581ae012b864d7c9088f0767fc8
artbio
parents:
diff changeset
30 help="Maximum overlap analyzed (nt) - must be an integer")
a35e6f9c1d34 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_signatures commit 6719543c5017d581ae012b864d7c9088f0767fc8
artbio
parents:
diff changeset
31 the_parser.add_argument(
a35e6f9c1d34 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_signatures commit 6719543c5017d581ae012b864d7c9088f0767fc8
artbio
parents:
diff changeset
32 '--output_h', action="store", type=str,
a35e6f9c1d34 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_signatures commit 6719543c5017d581ae012b864d7c9088f0767fc8
artbio
parents:
diff changeset
33 help="h-signature dataframe")
a35e6f9c1d34 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_signatures commit 6719543c5017d581ae012b864d7c9088f0767fc8
artbio
parents:
diff changeset
34 the_parser.add_argument(
a35e6f9c1d34 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_signatures commit 6719543c5017d581ae012b864d7c9088f0767fc8
artbio
parents:
diff changeset
35 '--output_z', action="store", type=str,
a35e6f9c1d34 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_signatures commit 6719543c5017d581ae012b864d7c9088f0767fc8
artbio
parents:
diff changeset
36 help="z-signature dataframe")
a35e6f9c1d34 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_signatures commit 6719543c5017d581ae012b864d7c9088f0767fc8
artbio
parents:
diff changeset
37 args = the_parser.parse_args()
a35e6f9c1d34 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_signatures commit 6719543c5017d581ae012b864d7c9088f0767fc8
artbio
parents:
diff changeset
38 return args
a35e6f9c1d34 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_signatures commit 6719543c5017d581ae012b864d7c9088f0767fc8
artbio
parents:
diff changeset
39
a35e6f9c1d34 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_signatures commit 6719543c5017d581ae012b864d7c9088f0767fc8
artbio
parents:
diff changeset
40
a35e6f9c1d34 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_signatures commit 6719543c5017d581ae012b864d7c9088f0767fc8
artbio
parents:
diff changeset
41 class Map:
a35e6f9c1d34 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_signatures commit 6719543c5017d581ae012b864d7c9088f0767fc8
artbio
parents:
diff changeset
42
a35e6f9c1d34 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_signatures commit 6719543c5017d581ae012b864d7c9088f0767fc8
artbio
parents:
diff changeset
43 def __init__(self, bam_file):
a35e6f9c1d34 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_signatures commit 6719543c5017d581ae012b864d7c9088f0767fc8
artbio
parents:
diff changeset
44 self.bam_object = pysam.AlignmentFile(bam_file, 'rb')
a35e6f9c1d34 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_signatures commit 6719543c5017d581ae012b864d7c9088f0767fc8
artbio
parents:
diff changeset
45 self.chromosomes = dict(zip(self.bam_object.references,
a35e6f9c1d34 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_signatures commit 6719543c5017d581ae012b864d7c9088f0767fc8
artbio
parents:
diff changeset
46 self.bam_object.lengths))
a35e6f9c1d34 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_signatures commit 6719543c5017d581ae012b864d7c9088f0767fc8
artbio
parents:
diff changeset
47 self.map_dict = self.create_map(self.bam_object)
a35e6f9c1d34 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_signatures commit 6719543c5017d581ae012b864d7c9088f0767fc8
artbio
parents:
diff changeset
48
a35e6f9c1d34 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_signatures commit 6719543c5017d581ae012b864d7c9088f0767fc8
artbio
parents:
diff changeset
49 def create_map(self, bam_object):
a35e6f9c1d34 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_signatures commit 6719543c5017d581ae012b864d7c9088f0767fc8
artbio
parents:
diff changeset
50 '''
a35e6f9c1d34 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_signatures commit 6719543c5017d581ae012b864d7c9088f0767fc8
artbio
parents:
diff changeset
51 Returns a map_dictionary {(chromosome,read_position,polarity):
a35e6f9c1d34 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_signatures commit 6719543c5017d581ae012b864d7c9088f0767fc8
artbio
parents:
diff changeset
52 [read_length, ...]}
a35e6f9c1d34 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_signatures commit 6719543c5017d581ae012b864d7c9088f0767fc8
artbio
parents:
diff changeset
53 '''
a35e6f9c1d34 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_signatures commit 6719543c5017d581ae012b864d7c9088f0767fc8
artbio
parents:
diff changeset
54 map_dictionary = defaultdict(list)
a35e6f9c1d34 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_signatures commit 6719543c5017d581ae012b864d7c9088f0767fc8
artbio
parents:
diff changeset
55 # get empty value for start and end of each chromosome
a35e6f9c1d34 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_signatures commit 6719543c5017d581ae012b864d7c9088f0767fc8
artbio
parents:
diff changeset
56 for chrom in self.chromosomes:
a35e6f9c1d34 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_signatures commit 6719543c5017d581ae012b864d7c9088f0767fc8
artbio
parents:
diff changeset
57 map_dictionary[(chrom, 1, 'F')] = []
a35e6f9c1d34 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_signatures commit 6719543c5017d581ae012b864d7c9088f0767fc8
artbio
parents:
diff changeset
58 map_dictionary[(chrom, self.chromosomes[chrom], 'F')] = []
a35e6f9c1d34 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_signatures commit 6719543c5017d581ae012b864d7c9088f0767fc8
artbio
parents:
diff changeset
59 for chrom in self.chromosomes:
a35e6f9c1d34 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_signatures commit 6719543c5017d581ae012b864d7c9088f0767fc8
artbio
parents:
diff changeset
60 for read in bam_object.fetch(chrom):
a35e6f9c1d34 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_signatures commit 6719543c5017d581ae012b864d7c9088f0767fc8
artbio
parents:
diff changeset
61 positions = read.positions # a list of covered positions
a35e6f9c1d34 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_signatures commit 6719543c5017d581ae012b864d7c9088f0767fc8
artbio
parents:
diff changeset
62 if read.is_reverse:
a35e6f9c1d34 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_signatures commit 6719543c5017d581ae012b864d7c9088f0767fc8
artbio
parents:
diff changeset
63 map_dictionary[(chrom, positions[-1]+1,
a35e6f9c1d34 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_signatures commit 6719543c5017d581ae012b864d7c9088f0767fc8
artbio
parents:
diff changeset
64 'R')].append(read.query_alignment_length)
a35e6f9c1d34 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_signatures commit 6719543c5017d581ae012b864d7c9088f0767fc8
artbio
parents:
diff changeset
65 else:
a35e6f9c1d34 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_signatures commit 6719543c5017d581ae012b864d7c9088f0767fc8
artbio
parents:
diff changeset
66 map_dictionary[(chrom, positions[0]+1,
a35e6f9c1d34 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_signatures commit 6719543c5017d581ae012b864d7c9088f0767fc8
artbio
parents:
diff changeset
67 'F')].append(read.query_alignment_length)
a35e6f9c1d34 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_signatures commit 6719543c5017d581ae012b864d7c9088f0767fc8
artbio
parents:
diff changeset
68 return map_dictionary
a35e6f9c1d34 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_signatures commit 6719543c5017d581ae012b864d7c9088f0767fc8
artbio
parents:
diff changeset
69
a35e6f9c1d34 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_signatures commit 6719543c5017d581ae012b864d7c9088f0767fc8
artbio
parents:
diff changeset
70 def signature_tables(self, minquery, maxquery, mintarget, maxtarget):
a35e6f9c1d34 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_signatures commit 6719543c5017d581ae012b864d7c9088f0767fc8
artbio
parents:
diff changeset
71 query_range = range(minquery, maxquery + 1)
a35e6f9c1d34 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_signatures commit 6719543c5017d581ae012b864d7c9088f0767fc8
artbio
parents:
diff changeset
72 target_range = range(mintarget, maxtarget + 1)
a35e6f9c1d34 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_signatures commit 6719543c5017d581ae012b864d7c9088f0767fc8
artbio
parents:
diff changeset
73 Query_table = defaultdict(dict)
a35e6f9c1d34 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_signatures commit 6719543c5017d581ae012b864d7c9088f0767fc8
artbio
parents:
diff changeset
74 Target_table = defaultdict(dict)
a35e6f9c1d34 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_signatures commit 6719543c5017d581ae012b864d7c9088f0767fc8
artbio
parents:
diff changeset
75 for key in self.map_dict:
a35e6f9c1d34 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_signatures commit 6719543c5017d581ae012b864d7c9088f0767fc8
artbio
parents:
diff changeset
76 for size in self.map_dict[key]:
a35e6f9c1d34 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_signatures commit 6719543c5017d581ae012b864d7c9088f0767fc8
artbio
parents:
diff changeset
77 if size in query_range or size in target_range:
a35e6f9c1d34 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_signatures commit 6719543c5017d581ae012b864d7c9088f0767fc8
artbio
parents:
diff changeset
78 if key[2] == 'F':
a35e6f9c1d34 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_signatures commit 6719543c5017d581ae012b864d7c9088f0767fc8
artbio
parents:
diff changeset
79 coordinate = key[1]
a35e6f9c1d34 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_signatures commit 6719543c5017d581ae012b864d7c9088f0767fc8
artbio
parents:
diff changeset
80 else:
a35e6f9c1d34 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_signatures commit 6719543c5017d581ae012b864d7c9088f0767fc8
artbio
parents:
diff changeset
81 coordinate = -key[1]
a35e6f9c1d34 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_signatures commit 6719543c5017d581ae012b864d7c9088f0767fc8
artbio
parents:
diff changeset
82 if size in query_range:
a35e6f9c1d34 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_signatures commit 6719543c5017d581ae012b864d7c9088f0767fc8
artbio
parents:
diff changeset
83 Query_table[key[0]][coordinate] = Query_table[key[0]].get(
a35e6f9c1d34 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_signatures commit 6719543c5017d581ae012b864d7c9088f0767fc8
artbio
parents:
diff changeset
84 coordinate, 0) + 1
a35e6f9c1d34 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_signatures commit 6719543c5017d581ae012b864d7c9088f0767fc8
artbio
parents:
diff changeset
85 if size in target_range:
a35e6f9c1d34 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_signatures commit 6719543c5017d581ae012b864d7c9088f0767fc8
artbio
parents:
diff changeset
86 Target_table[key[0]][coordinate] = \
a35e6f9c1d34 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_signatures commit 6719543c5017d581ae012b864d7c9088f0767fc8
artbio
parents:
diff changeset
87 Target_table[key[0]].get(coordinate, 0) + 1
a35e6f9c1d34 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_signatures commit 6719543c5017d581ae012b864d7c9088f0767fc8
artbio
parents:
diff changeset
88 return Query_table, Target_table
a35e6f9c1d34 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_signatures commit 6719543c5017d581ae012b864d7c9088f0767fc8
artbio
parents:
diff changeset
89
a35e6f9c1d34 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_signatures commit 6719543c5017d581ae012b864d7c9088f0767fc8
artbio
parents:
diff changeset
90 def compute_signature_z(self, minquery, maxquery, mintarget, maxtarget,
a35e6f9c1d34 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_signatures commit 6719543c5017d581ae012b864d7c9088f0767fc8
artbio
parents:
diff changeset
91 scope, zscore="no"):
a35e6f9c1d34 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_signatures commit 6719543c5017d581ae012b864d7c9088f0767fc8
artbio
parents:
diff changeset
92 Query_table, Target_table = self.signature_tables(minquery, maxquery,
a35e6f9c1d34 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_signatures commit 6719543c5017d581ae012b864d7c9088f0767fc8
artbio
parents:
diff changeset
93 mintarget, maxtarget)
a35e6f9c1d34 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_signatures commit 6719543c5017d581ae012b864d7c9088f0767fc8
artbio
parents:
diff changeset
94 frequency_table = defaultdict(dict)
a35e6f9c1d34 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_signatures commit 6719543c5017d581ae012b864d7c9088f0767fc8
artbio
parents:
diff changeset
95 for chrom in self.chromosomes:
a35e6f9c1d34 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_signatures commit 6719543c5017d581ae012b864d7c9088f0767fc8
artbio
parents:
diff changeset
96 for overlap in scope:
a35e6f9c1d34 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_signatures commit 6719543c5017d581ae012b864d7c9088f0767fc8
artbio
parents:
diff changeset
97 frequency_table[chrom][overlap] = 0
a35e6f9c1d34 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_signatures commit 6719543c5017d581ae012b864d7c9088f0767fc8
artbio
parents:
diff changeset
98 for chrom in Query_table:
a35e6f9c1d34 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_signatures commit 6719543c5017d581ae012b864d7c9088f0767fc8
artbio
parents:
diff changeset
99 for coord in Query_table[chrom]:
a35e6f9c1d34 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_signatures commit 6719543c5017d581ae012b864d7c9088f0767fc8
artbio
parents:
diff changeset
100 for overlap in scope:
a35e6f9c1d34 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_signatures commit 6719543c5017d581ae012b864d7c9088f0767fc8
artbio
parents:
diff changeset
101 frequency_table[chrom][overlap] += min(
a35e6f9c1d34 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_signatures commit 6719543c5017d581ae012b864d7c9088f0767fc8
artbio
parents:
diff changeset
102 Query_table[chrom][coord],
a35e6f9c1d34 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_signatures commit 6719543c5017d581ae012b864d7c9088f0767fc8
artbio
parents:
diff changeset
103 Target_table[chrom].get(-coord - overlap + 1, 0))
a35e6f9c1d34 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_signatures commit 6719543c5017d581ae012b864d7c9088f0767fc8
artbio
parents:
diff changeset
104 # since we want the number of pairs, not the number or paired reads
a35e6f9c1d34 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_signatures commit 6719543c5017d581ae012b864d7c9088f0767fc8
artbio
parents:
diff changeset
105 # to do: what in complex cases
a35e6f9c1d34 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_signatures commit 6719543c5017d581ae012b864d7c9088f0767fc8
artbio
parents:
diff changeset
106 # with query and target sizes partially overlap ?
a35e6f9c1d34 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_signatures commit 6719543c5017d581ae012b864d7c9088f0767fc8
artbio
parents:
diff changeset
107 for chrom in frequency_table:
a35e6f9c1d34 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_signatures commit 6719543c5017d581ae012b864d7c9088f0767fc8
artbio
parents:
diff changeset
108 for overlap in frequency_table[chrom]:
a35e6f9c1d34 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_signatures commit 6719543c5017d581ae012b864d7c9088f0767fc8
artbio
parents:
diff changeset
109 frequency_table[chrom][overlap] /= 2
a35e6f9c1d34 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_signatures commit 6719543c5017d581ae012b864d7c9088f0767fc8
artbio
parents:
diff changeset
110 # compute overlaps for all chromosomes merged
a35e6f9c1d34 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_signatures commit 6719543c5017d581ae012b864d7c9088f0767fc8
artbio
parents:
diff changeset
111 for overlap in scope:
a35e6f9c1d34 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_signatures commit 6719543c5017d581ae012b864d7c9088f0767fc8
artbio
parents:
diff changeset
112 accumulator = []
a35e6f9c1d34 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_signatures commit 6719543c5017d581ae012b864d7c9088f0767fc8
artbio
parents:
diff changeset
113 for chrom in frequency_table:
a35e6f9c1d34 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_signatures commit 6719543c5017d581ae012b864d7c9088f0767fc8
artbio
parents:
diff changeset
114 if chrom != 'all_chromosomes':
a35e6f9c1d34 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_signatures commit 6719543c5017d581ae012b864d7c9088f0767fc8
artbio
parents:
diff changeset
115 accumulator.append(frequency_table[chrom][overlap])
a35e6f9c1d34 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_signatures commit 6719543c5017d581ae012b864d7c9088f0767fc8
artbio
parents:
diff changeset
116 frequency_table['all_chromosomes'][overlap] = sum(accumulator)
a35e6f9c1d34 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_signatures commit 6719543c5017d581ae012b864d7c9088f0767fc8
artbio
parents:
diff changeset
117 return self.stringify_table(frequency_table)
a35e6f9c1d34 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_signatures commit 6719543c5017d581ae012b864d7c9088f0767fc8
artbio
parents:
diff changeset
118
a35e6f9c1d34 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_signatures commit 6719543c5017d581ae012b864d7c9088f0767fc8
artbio
parents:
diff changeset
119 def compute_signature_h(self, minquery, maxquery, mintarget,
a35e6f9c1d34 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_signatures commit 6719543c5017d581ae012b864d7c9088f0767fc8
artbio
parents:
diff changeset
120 maxtarget, scope):
a35e6f9c1d34 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_signatures commit 6719543c5017d581ae012b864d7c9088f0767fc8
artbio
parents:
diff changeset
121 Query_table, Target_table = self.signature_tables(minquery, maxquery,
a35e6f9c1d34 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_signatures commit 6719543c5017d581ae012b864d7c9088f0767fc8
artbio
parents:
diff changeset
122 mintarget, maxtarget)
a35e6f9c1d34 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_signatures commit 6719543c5017d581ae012b864d7c9088f0767fc8
artbio
parents:
diff changeset
123 frequency_table = defaultdict(dict)
a35e6f9c1d34 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_signatures commit 6719543c5017d581ae012b864d7c9088f0767fc8
artbio
parents:
diff changeset
124 for chrom in self.chromosomes:
a35e6f9c1d34 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_signatures commit 6719543c5017d581ae012b864d7c9088f0767fc8
artbio
parents:
diff changeset
125 for overlap in scope:
a35e6f9c1d34 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_signatures commit 6719543c5017d581ae012b864d7c9088f0767fc8
artbio
parents:
diff changeset
126 frequency_table[chrom][overlap] = 0
a35e6f9c1d34 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_signatures commit 6719543c5017d581ae012b864d7c9088f0767fc8
artbio
parents:
diff changeset
127 for chrom in Query_table:
a35e6f9c1d34 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_signatures commit 6719543c5017d581ae012b864d7c9088f0767fc8
artbio
parents:
diff changeset
128 Total_Query_Numb = 0
a35e6f9c1d34 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_signatures commit 6719543c5017d581ae012b864d7c9088f0767fc8
artbio
parents:
diff changeset
129 for coord in Query_table[chrom]:
a35e6f9c1d34 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_signatures commit 6719543c5017d581ae012b864d7c9088f0767fc8
artbio
parents:
diff changeset
130 Total_Query_Numb += Query_table[chrom][coord]
a35e6f9c1d34 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_signatures commit 6719543c5017d581ae012b864d7c9088f0767fc8
artbio
parents:
diff changeset
131 for coord in Query_table[chrom]:
a35e6f9c1d34 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_signatures commit 6719543c5017d581ae012b864d7c9088f0767fc8
artbio
parents:
diff changeset
132 local_table = dict([(overlap, 0) for overlap in scope])
a35e6f9c1d34 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_signatures commit 6719543c5017d581ae012b864d7c9088f0767fc8
artbio
parents:
diff changeset
133 number_of_targets = 0
a35e6f9c1d34 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_signatures commit 6719543c5017d581ae012b864d7c9088f0767fc8
artbio
parents:
diff changeset
134 for overlap in scope:
a35e6f9c1d34 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_signatures commit 6719543c5017d581ae012b864d7c9088f0767fc8
artbio
parents:
diff changeset
135 local_table[overlap] += Query_table[chrom][coord] * \
a35e6f9c1d34 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_signatures commit 6719543c5017d581ae012b864d7c9088f0767fc8
artbio
parents:
diff changeset
136 Target_table[chrom].get(-coord - overlap + 1, 0)
a35e6f9c1d34 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_signatures commit 6719543c5017d581ae012b864d7c9088f0767fc8
artbio
parents:
diff changeset
137 number_of_targets += Target_table[chrom].get(
a35e6f9c1d34 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_signatures commit 6719543c5017d581ae012b864d7c9088f0767fc8
artbio
parents:
diff changeset
138 -coord - overlap + 1, 0)
a35e6f9c1d34 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_signatures commit 6719543c5017d581ae012b864d7c9088f0767fc8
artbio
parents:
diff changeset
139 for overlap in scope:
a35e6f9c1d34 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_signatures commit 6719543c5017d581ae012b864d7c9088f0767fc8
artbio
parents:
diff changeset
140 try:
a35e6f9c1d34 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_signatures commit 6719543c5017d581ae012b864d7c9088f0767fc8
artbio
parents:
diff changeset
141 frequency_table[chrom][overlap] += \
a35e6f9c1d34 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_signatures commit 6719543c5017d581ae012b864d7c9088f0767fc8
artbio
parents:
diff changeset
142 local_table[overlap] / number_of_targets \
a35e6f9c1d34 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_signatures commit 6719543c5017d581ae012b864d7c9088f0767fc8
artbio
parents:
diff changeset
143 / float(Total_Query_Numb)
a35e6f9c1d34 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_signatures commit 6719543c5017d581ae012b864d7c9088f0767fc8
artbio
parents:
diff changeset
144 except ZeroDivisionError:
a35e6f9c1d34 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_signatures commit 6719543c5017d581ae012b864d7c9088f0767fc8
artbio
parents:
diff changeset
145 continue
a35e6f9c1d34 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_signatures commit 6719543c5017d581ae012b864d7c9088f0767fc8
artbio
parents:
diff changeset
146 # compute overlap probabilities for all chromosomes merged
a35e6f9c1d34 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_signatures commit 6719543c5017d581ae012b864d7c9088f0767fc8
artbio
parents:
diff changeset
147 general_frequency_table = dict([(overlap, 0) for overlap in scope])
a35e6f9c1d34 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_signatures commit 6719543c5017d581ae012b864d7c9088f0767fc8
artbio
parents:
diff changeset
148 total_aligned_reads = 0
a35e6f9c1d34 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_signatures commit 6719543c5017d581ae012b864d7c9088f0767fc8
artbio
parents:
diff changeset
149 for chrom in frequency_table:
a35e6f9c1d34 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_signatures commit 6719543c5017d581ae012b864d7c9088f0767fc8
artbio
parents:
diff changeset
150 for overlap in frequency_table[chrom]:
a35e6f9c1d34 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_signatures commit 6719543c5017d581ae012b864d7c9088f0767fc8
artbio
parents:
diff changeset
151 total_aligned_reads += self.bam_object.count(chrom)
a35e6f9c1d34 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_signatures commit 6719543c5017d581ae012b864d7c9088f0767fc8
artbio
parents:
diff changeset
152 for chrom in frequency_table:
a35e6f9c1d34 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_signatures commit 6719543c5017d581ae012b864d7c9088f0767fc8
artbio
parents:
diff changeset
153 for overlap in frequency_table[chrom]:
a35e6f9c1d34 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_signatures commit 6719543c5017d581ae012b864d7c9088f0767fc8
artbio
parents:
diff changeset
154 try:
a35e6f9c1d34 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_signatures commit 6719543c5017d581ae012b864d7c9088f0767fc8
artbio
parents:
diff changeset
155 general_frequency_table[overlap] += \
a35e6f9c1d34 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_signatures commit 6719543c5017d581ae012b864d7c9088f0767fc8
artbio
parents:
diff changeset
156 frequency_table[chrom][overlap] / total_aligned_reads \
a35e6f9c1d34 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_signatures commit 6719543c5017d581ae012b864d7c9088f0767fc8
artbio
parents:
diff changeset
157 * self.bam_object.count(chrom)
a35e6f9c1d34 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_signatures commit 6719543c5017d581ae012b864d7c9088f0767fc8
artbio
parents:
diff changeset
158 except ZeroDivisionError:
a35e6f9c1d34 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_signatures commit 6719543c5017d581ae012b864d7c9088f0767fc8
artbio
parents:
diff changeset
159 continue
a35e6f9c1d34 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_signatures commit 6719543c5017d581ae012b864d7c9088f0767fc8
artbio
parents:
diff changeset
160 for overlap in general_frequency_table:
a35e6f9c1d34 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_signatures commit 6719543c5017d581ae012b864d7c9088f0767fc8
artbio
parents:
diff changeset
161 frequency_table['all_chromosomes'][overlap] = \
a35e6f9c1d34 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_signatures commit 6719543c5017d581ae012b864d7c9088f0767fc8
artbio
parents:
diff changeset
162 general_frequency_table[overlap]
a35e6f9c1d34 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_signatures commit 6719543c5017d581ae012b864d7c9088f0767fc8
artbio
parents:
diff changeset
163 return self.stringify_table(frequency_table)
a35e6f9c1d34 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_signatures commit 6719543c5017d581ae012b864d7c9088f0767fc8
artbio
parents:
diff changeset
164
a35e6f9c1d34 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_signatures commit 6719543c5017d581ae012b864d7c9088f0767fc8
artbio
parents:
diff changeset
165 def stringify_table(self, frequency_table):
a35e6f9c1d34 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_signatures commit 6719543c5017d581ae012b864d7c9088f0767fc8
artbio
parents:
diff changeset
166 '''
a35e6f9c1d34 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_signatures commit 6719543c5017d581ae012b864d7c9088f0767fc8
artbio
parents:
diff changeset
167 method both to compute z-score and to return a writable string
a35e6f9c1d34 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_signatures commit 6719543c5017d581ae012b864d7c9088f0767fc8
artbio
parents:
diff changeset
168 '''
a35e6f9c1d34 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_signatures commit 6719543c5017d581ae012b864d7c9088f0767fc8
artbio
parents:
diff changeset
169 tablestring = []
a35e6f9c1d34 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_signatures commit 6719543c5017d581ae012b864d7c9088f0767fc8
artbio
parents:
diff changeset
170 for chrom in sorted(frequency_table):
a35e6f9c1d34 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_signatures commit 6719543c5017d581ae012b864d7c9088f0767fc8
artbio
parents:
diff changeset
171 accumulator = []
a35e6f9c1d34 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_signatures commit 6719543c5017d581ae012b864d7c9088f0767fc8
artbio
parents:
diff changeset
172 for overlap in frequency_table[chrom]:
a35e6f9c1d34 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_signatures commit 6719543c5017d581ae012b864d7c9088f0767fc8
artbio
parents:
diff changeset
173 accumulator.append(frequency_table[chrom][overlap])
a35e6f9c1d34 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_signatures commit 6719543c5017d581ae012b864d7c9088f0767fc8
artbio
parents:
diff changeset
174 z_mean = numpy.mean(accumulator)
a35e6f9c1d34 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_signatures commit 6719543c5017d581ae012b864d7c9088f0767fc8
artbio
parents:
diff changeset
175 z_std = numpy.std(accumulator)
a35e6f9c1d34 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_signatures commit 6719543c5017d581ae012b864d7c9088f0767fc8
artbio
parents:
diff changeset
176 if z_std == 0:
a35e6f9c1d34 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_signatures commit 6719543c5017d581ae012b864d7c9088f0767fc8
artbio
parents:
diff changeset
177 for overlap in sorted(frequency_table[chrom]):
a35e6f9c1d34 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_signatures commit 6719543c5017d581ae012b864d7c9088f0767fc8
artbio
parents:
diff changeset
178 tablestring.append('%s\t%s\t%s\t%s\n' % (
a35e6f9c1d34 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_signatures commit 6719543c5017d581ae012b864d7c9088f0767fc8
artbio
parents:
diff changeset
179 chrom, str(overlap),
a35e6f9c1d34 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_signatures commit 6719543c5017d581ae012b864d7c9088f0767fc8
artbio
parents:
diff changeset
180 str(frequency_table[chrom][overlap]), str(0)))
a35e6f9c1d34 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_signatures commit 6719543c5017d581ae012b864d7c9088f0767fc8
artbio
parents:
diff changeset
181 else:
a35e6f9c1d34 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_signatures commit 6719543c5017d581ae012b864d7c9088f0767fc8
artbio
parents:
diff changeset
182 for overlap in sorted(frequency_table[chrom]):
a35e6f9c1d34 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_signatures commit 6719543c5017d581ae012b864d7c9088f0767fc8
artbio
parents:
diff changeset
183 tablestring.append('%s\t%s\t%s\t%s\n' % (
a35e6f9c1d34 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_signatures commit 6719543c5017d581ae012b864d7c9088f0767fc8
artbio
parents:
diff changeset
184 chrom, str(overlap),
a35e6f9c1d34 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_signatures commit 6719543c5017d581ae012b864d7c9088f0767fc8
artbio
parents:
diff changeset
185 str(frequency_table[chrom][overlap]),
a35e6f9c1d34 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_signatures commit 6719543c5017d581ae012b864d7c9088f0767fc8
artbio
parents:
diff changeset
186 str((frequency_table[chrom][overlap] - z_mean)/z_std)))
a35e6f9c1d34 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_signatures commit 6719543c5017d581ae012b864d7c9088f0767fc8
artbio
parents:
diff changeset
187 return ''.join(tablestring)
a35e6f9c1d34 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_signatures commit 6719543c5017d581ae012b864d7c9088f0767fc8
artbio
parents:
diff changeset
188
a35e6f9c1d34 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_signatures commit 6719543c5017d581ae012b864d7c9088f0767fc8
artbio
parents:
diff changeset
189
a35e6f9c1d34 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_signatures commit 6719543c5017d581ae012b864d7c9088f0767fc8
artbio
parents:
diff changeset
190
a35e6f9c1d34 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_signatures commit 6719543c5017d581ae012b864d7c9088f0767fc8
artbio
parents:
diff changeset
191 def main(input, minquery, maxquery, mintarget, maxtarget, minscope, maxscope,
a35e6f9c1d34 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_signatures commit 6719543c5017d581ae012b864d7c9088f0767fc8
artbio
parents:
diff changeset
192 output_h, output_z, genome_wide=False, zscore="no"):
a35e6f9c1d34 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_signatures commit 6719543c5017d581ae012b864d7c9088f0767fc8
artbio
parents:
diff changeset
193 H = open(output_h, 'w')
a35e6f9c1d34 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_signatures commit 6719543c5017d581ae012b864d7c9088f0767fc8
artbio
parents:
diff changeset
194 Z = open(output_z, 'w')
a35e6f9c1d34 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_signatures commit 6719543c5017d581ae012b864d7c9088f0767fc8
artbio
parents:
diff changeset
195 mapobj = Map(input)
a35e6f9c1d34 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_signatures commit 6719543c5017d581ae012b864d7c9088f0767fc8
artbio
parents:
diff changeset
196 scope = range(minscope, maxscope + 1)
a35e6f9c1d34 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_signatures commit 6719543c5017d581ae012b864d7c9088f0767fc8
artbio
parents:
diff changeset
197 Z.write(mapobj.compute_signature_z(minquery, maxquery, mintarget,
a35e6f9c1d34 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_signatures commit 6719543c5017d581ae012b864d7c9088f0767fc8
artbio
parents:
diff changeset
198 maxtarget, scope, zscore="no"))
a35e6f9c1d34 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_signatures commit 6719543c5017d581ae012b864d7c9088f0767fc8
artbio
parents:
diff changeset
199 H.write(mapobj.compute_signature_h(minquery, maxquery, mintarget,
a35e6f9c1d34 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_signatures commit 6719543c5017d581ae012b864d7c9088f0767fc8
artbio
parents:
diff changeset
200 maxtarget, scope))
a35e6f9c1d34 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_signatures commit 6719543c5017d581ae012b864d7c9088f0767fc8
artbio
parents:
diff changeset
201 H.close()
a35e6f9c1d34 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_signatures commit 6719543c5017d581ae012b864d7c9088f0767fc8
artbio
parents:
diff changeset
202 Z.close()
a35e6f9c1d34 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_signatures commit 6719543c5017d581ae012b864d7c9088f0767fc8
artbio
parents:
diff changeset
203
a35e6f9c1d34 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_signatures commit 6719543c5017d581ae012b864d7c9088f0767fc8
artbio
parents:
diff changeset
204
a35e6f9c1d34 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_signatures commit 6719543c5017d581ae012b864d7c9088f0767fc8
artbio
parents:
diff changeset
205 if __name__ == "__main__":
a35e6f9c1d34 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_signatures commit 6719543c5017d581ae012b864d7c9088f0767fc8
artbio
parents:
diff changeset
206 args = Parser()
a35e6f9c1d34 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_signatures commit 6719543c5017d581ae012b864d7c9088f0767fc8
artbio
parents:
diff changeset
207 main(args.input, args.minquery, args.maxquery, args.mintarget,
a35e6f9c1d34 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_signatures commit 6719543c5017d581ae012b864d7c9088f0767fc8
artbio
parents:
diff changeset
208 args.maxtarget, args.minscope, args.maxscope, args.output_h,
a35e6f9c1d34 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_signatures commit 6719543c5017d581ae012b864d7c9088f0767fc8
artbio
parents:
diff changeset
209 args.output_z)