Mercurial > repos > artbio > small_rna_maps

annotate small_rna_maps.py @ 17:b28dcd4051e8 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_maps commit 16f15e5ab2b79590a8ae410f76434aa6690c1fc4
author artbio
date Thu, 15 Nov 2018 12:29:57 -0500
parents 600e2498bd21
children 2c95c899d0a4
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
rev   line source
0
6d48150495e3 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_maps commit d4d8106d66b65679a1a685ab94bfcf99cdb7b959
artbio
parents:
diff changeset
1 import argparse
6d48150495e3 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_maps commit d4d8106d66b65679a1a685ab94bfcf99cdb7b959
artbio
parents:
diff changeset
2 from collections import defaultdict
6d48150495e3 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_maps commit d4d8106d66b65679a1a685ab94bfcf99cdb7b959
artbio
parents:
diff changeset
3
6d48150495e3 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_maps commit d4d8106d66b65679a1a685ab94bfcf99cdb7b959
artbio
parents:
diff changeset
4 import numpy
6d48150495e3 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_maps commit d4d8106d66b65679a1a685ab94bfcf99cdb7b959
artbio
parents:
diff changeset
5
6d48150495e3 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_maps commit d4d8106d66b65679a1a685ab94bfcf99cdb7b959
artbio
parents:
diff changeset
6 import pysam
6d48150495e3 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_maps commit d4d8106d66b65679a1a685ab94bfcf99cdb7b959
artbio
parents:
diff changeset
7
6d48150495e3 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_maps commit d4d8106d66b65679a1a685ab94bfcf99cdb7b959
artbio
parents:
diff changeset
8
6d48150495e3 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_maps commit d4d8106d66b65679a1a685ab94bfcf99cdb7b959
artbio
parents:
diff changeset
9 def Parser():
6d48150495e3 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_maps commit d4d8106d66b65679a1a685ab94bfcf99cdb7b959
artbio
parents:
diff changeset
10 the_parser = argparse.ArgumentParser()
2
507383cce5a8 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_maps commit edbb53cb13b52bf8e71c562fa8acc2c3be2fb270
artbio
parents: 1
diff changeset
11 the_parser.add_argument('--inputs', dest='inputs', required=True,
507383cce5a8 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_maps commit edbb53cb13b52bf8e71c562fa8acc2c3be2fb270
artbio
parents: 1
diff changeset
12 nargs='+', help='list of input BAM files')
8
1827b74f872b planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_maps commit e4588eb6c329e4516e9bcfa084a383be81b55c60
artbio
parents: 5
diff changeset
13 the_parser.add_argument('--minsize', dest='minsize', type=int,
1827b74f872b planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_maps commit e4588eb6c329e4516e9bcfa084a383be81b55c60
artbio
parents: 5
diff changeset
14 default=0, help='minimal size of reads')
1827b74f872b planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_maps commit e4588eb6c329e4516e9bcfa084a383be81b55c60
artbio
parents: 5
diff changeset
15 the_parser.add_argument('--maxsize', dest='maxsize', type=int,
1827b74f872b planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_maps commit e4588eb6c329e4516e9bcfa084a383be81b55c60
artbio
parents: 5
diff changeset
16 default=10000, help='maximal size of reads')
9
3ea75c573429 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_maps commit 6199193c7fe2cb56403eea8af0b40d44f7311fd5
artbio
parents: 8
diff changeset
17 the_parser.add_argument('--cluster', dest='cluster', type=int,
3ea75c573429 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_maps commit 6199193c7fe2cb56403eea8af0b40d44f7311fd5
artbio
parents: 8
diff changeset
18 default=0, help='clustering distance')
2
507383cce5a8 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_maps commit edbb53cb13b52bf8e71c562fa8acc2c3be2fb270
artbio
parents: 1
diff changeset
19 the_parser.add_argument('--sample_names', dest='sample_names',
507383cce5a8 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_maps commit edbb53cb13b52bf8e71c562fa8acc2c3be2fb270
artbio
parents: 1
diff changeset
20 required=True, nargs='+',
507383cce5a8 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_maps commit edbb53cb13b52bf8e71c562fa8acc2c3be2fb270
artbio
parents: 1
diff changeset
21 help='list of sample names')
507383cce5a8 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_maps commit edbb53cb13b52bf8e71c562fa8acc2c3be2fb270
artbio
parents: 1
diff changeset
22 the_parser.add_argument('--outputs', nargs='+', action='store',
507383cce5a8 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_maps commit edbb53cb13b52bf8e71c562fa8acc2c3be2fb270
artbio
parents: 1
diff changeset
23 help='list of two output paths (only two)')
507383cce5a8 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_maps commit edbb53cb13b52bf8e71c562fa8acc2c3be2fb270
artbio
parents: 1
diff changeset
24 the_parser.add_argument('-M', '--plot_methods', nargs='+', action='store',
507383cce5a8 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_maps commit edbb53cb13b52bf8e71c562fa8acc2c3be2fb270
artbio
parents: 1
diff changeset
25 help='list of 2 plot methods (only two) among:\
507383cce5a8 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_maps commit edbb53cb13b52bf8e71c562fa8acc2c3be2fb270
artbio
parents: 1
diff changeset
26 Counts, Max, Mean, Median, Coverage and Size')
17
b28dcd4051e8 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_maps commit 16f15e5ab2b79590a8ae410f76434aa6690c1fc4
artbio
parents: 16
diff changeset
27 the_parser.add_argument('--nostrand', action='store_true',
b28dcd4051e8 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_maps commit 16f15e5ab2b79590a8ae410f76434aa6690c1fc4
artbio
parents: 16
diff changeset
28 help='Consider reads regardless their polarity')
b28dcd4051e8 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_maps commit 16f15e5ab2b79590a8ae410f76434aa6690c1fc4
artbio
parents: 16
diff changeset
29
0
6d48150495e3 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_maps commit d4d8106d66b65679a1a685ab94bfcf99cdb7b959
artbio
parents:
diff changeset
30 args = the_parser.parse_args()
6d48150495e3 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_maps commit d4d8106d66b65679a1a685ab94bfcf99cdb7b959
artbio
parents:
diff changeset
31 return args
6d48150495e3 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_maps commit d4d8106d66b65679a1a685ab94bfcf99cdb7b959
artbio
parents:
diff changeset
32
6d48150495e3 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_maps commit d4d8106d66b65679a1a685ab94bfcf99cdb7b959
artbio
parents:
diff changeset
33
6d48150495e3 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_maps commit d4d8106d66b65679a1a685ab94bfcf99cdb7b959
artbio
parents:
diff changeset
34 class Map:
6d48150495e3 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_maps commit d4d8106d66b65679a1a685ab94bfcf99cdb7b959
artbio
parents:
diff changeset
35
17
b28dcd4051e8 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_maps commit 16f15e5ab2b79590a8ae410f76434aa6690c1fc4
artbio
parents: 16
diff changeset
36 def __init__(self, bam_file, sample, minsize, maxsize, cluster, nostrand):
0
6d48150495e3 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_maps commit d4d8106d66b65679a1a685ab94bfcf99cdb7b959
artbio
parents:
diff changeset
37 self.sample_name = sample
8
1827b74f872b planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_maps commit e4588eb6c329e4516e9bcfa084a383be81b55c60
artbio
parents: 5
diff changeset
38 self.minsize = minsize
1827b74f872b planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_maps commit e4588eb6c329e4516e9bcfa084a383be81b55c60
artbio
parents: 5
diff changeset
39 self.maxsize = maxsize
9
3ea75c573429 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_maps commit 6199193c7fe2cb56403eea8af0b40d44f7311fd5
artbio
parents: 8
diff changeset
40 self.cluster = cluster
17
b28dcd4051e8 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_maps commit 16f15e5ab2b79590a8ae410f76434aa6690c1fc4
artbio
parents: 16
diff changeset
41 if not nostrand:
b28dcd4051e8 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_maps commit 16f15e5ab2b79590a8ae410f76434aa6690c1fc4
artbio
parents: 16
diff changeset
42 self.nostrand = False
b28dcd4051e8 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_maps commit 16f15e5ab2b79590a8ae410f76434aa6690c1fc4
artbio
parents: 16
diff changeset
43 else:
b28dcd4051e8 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_maps commit 16f15e5ab2b79590a8ae410f76434aa6690c1fc4
artbio
parents: 16
diff changeset
44 self.nostrand = True
0
6d48150495e3 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_maps commit d4d8106d66b65679a1a685ab94bfcf99cdb7b959
artbio
parents:
diff changeset
45 self.bam_object = pysam.AlignmentFile(bam_file, 'rb')
6d48150495e3 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_maps commit d4d8106d66b65679a1a685ab94bfcf99cdb7b959
artbio
parents:
diff changeset
46 self.chromosomes = dict(zip(self.bam_object.references,
6d48150495e3 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_maps commit d4d8106d66b65679a1a685ab94bfcf99cdb7b959
artbio
parents:
diff changeset
47 self.bam_object.lengths))
8
1827b74f872b planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_maps commit e4588eb6c329e4516e9bcfa084a383be81b55c60
artbio
parents: 5
diff changeset
48 self.map_dict = self.create_map(self.bam_object, self.minsize,
17
b28dcd4051e8 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_maps commit 16f15e5ab2b79590a8ae410f76434aa6690c1fc4
artbio
parents: 16
diff changeset
49 self.maxsize, self.nostrand)
9
3ea75c573429 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_maps commit 6199193c7fe2cb56403eea8af0b40d44f7311fd5
artbio
parents: 8
diff changeset
50 if self.cluster:
3ea75c573429 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_maps commit 6199193c7fe2cb56403eea8af0b40d44f7311fd5
artbio
parents: 8
diff changeset
51 self.map_dict = self.tile_map(self.map_dict, self.cluster)
0
6d48150495e3 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_maps commit d4d8106d66b65679a1a685ab94bfcf99cdb7b959
artbio
parents:
diff changeset
52
17
b28dcd4051e8 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_maps commit 16f15e5ab2b79590a8ae410f76434aa6690c1fc4
artbio
parents: 16
diff changeset
53 def create_map(self, bam_object, minsize, maxsize, nostrand=False):
0
6d48150495e3 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_maps commit d4d8106d66b65679a1a685ab94bfcf99cdb7b959
artbio
parents:
diff changeset
54 '''
6d48150495e3 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_maps commit d4d8106d66b65679a1a685ab94bfcf99cdb7b959
artbio
parents:
diff changeset
55 Returns a map_dictionary {(chromosome,read_position,polarity):
6d48150495e3 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_maps commit d4d8106d66b65679a1a685ab94bfcf99cdb7b959
artbio
parents:
diff changeset
56 [read_length, ...]}
6d48150495e3 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_maps commit d4d8106d66b65679a1a685ab94bfcf99cdb7b959
artbio
parents:
diff changeset
57 '''
6d48150495e3 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_maps commit d4d8106d66b65679a1a685ab94bfcf99cdb7b959
artbio
parents:
diff changeset
58 map_dictionary = defaultdict(list)
6d48150495e3 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_maps commit d4d8106d66b65679a1a685ab94bfcf99cdb7b959
artbio
parents:
diff changeset
59 for chrom in self.chromosomes:
9
3ea75c573429 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_maps commit 6199193c7fe2cb56403eea8af0b40d44f7311fd5
artbio
parents: 8
diff changeset
60 # get empty value for start and end of each chromosome
0
6d48150495e3 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_maps commit d4d8106d66b65679a1a685ab94bfcf99cdb7b959
artbio
parents:
diff changeset
61 map_dictionary[(chrom, 1, 'F')] = []
6d48150495e3 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_maps commit d4d8106d66b65679a1a685ab94bfcf99cdb7b959
artbio
parents:
diff changeset
62 map_dictionary[(chrom, self.chromosomes[chrom], 'F')] = []
17
b28dcd4051e8 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_maps commit 16f15e5ab2b79590a8ae410f76434aa6690c1fc4
artbio
parents: 16
diff changeset
63 if not nostrand:
b28dcd4051e8 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_maps commit 16f15e5ab2b79590a8ae410f76434aa6690c1fc4
artbio
parents: 16
diff changeset
64 for read in bam_object.fetch(chrom):
b28dcd4051e8 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_maps commit 16f15e5ab2b79590a8ae410f76434aa6690c1fc4
artbio
parents: 16
diff changeset
65 positions = read.positions # a list of covered positions
b28dcd4051e8 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_maps commit 16f15e5ab2b79590a8ae410f76434aa6690c1fc4
artbio
parents: 16
diff changeset
66 if read.is_reverse:
b28dcd4051e8 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_maps commit 16f15e5ab2b79590a8ae410f76434aa6690c1fc4
artbio
parents: 16
diff changeset
67 map_dictionary[(chrom, positions[-1]+1, 'R')].append(
b28dcd4051e8 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_maps commit 16f15e5ab2b79590a8ae410f76434aa6690c1fc4
artbio
parents: 16
diff changeset
68 read.query_alignment_length)
b28dcd4051e8 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_maps commit 16f15e5ab2b79590a8ae410f76434aa6690c1fc4
artbio
parents: 16
diff changeset
69 else:
b28dcd4051e8 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_maps commit 16f15e5ab2b79590a8ae410f76434aa6690c1fc4
artbio
parents: 16
diff changeset
70 map_dictionary[(chrom, positions[0]+1, 'F')].append(
b28dcd4051e8 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_maps commit 16f15e5ab2b79590a8ae410f76434aa6690c1fc4
artbio
parents: 16
diff changeset
71 read.query_alignment_length)
b28dcd4051e8 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_maps commit 16f15e5ab2b79590a8ae410f76434aa6690c1fc4
artbio
parents: 16
diff changeset
72 else:
b28dcd4051e8 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_maps commit 16f15e5ab2b79590a8ae410f76434aa6690c1fc4
artbio
parents: 16
diff changeset
73 for read in bam_object.fetch(chrom):
b28dcd4051e8 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_maps commit 16f15e5ab2b79590a8ae410f76434aa6690c1fc4
artbio
parents: 16
diff changeset
74 positions = read.positions # a list of covered positions
b28dcd4051e8 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_maps commit 16f15e5ab2b79590a8ae410f76434aa6690c1fc4
artbio
parents: 16
diff changeset
75 if read.is_reverse:
b28dcd4051e8 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_maps commit 16f15e5ab2b79590a8ae410f76434aa6690c1fc4
artbio
parents: 16
diff changeset
76 map_dictionary[(chrom, positions[-1]+1, 'F')].append(
b28dcd4051e8 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_maps commit 16f15e5ab2b79590a8ae410f76434aa6690c1fc4
artbio
parents: 16
diff changeset
77 read.query_alignment_length)
b28dcd4051e8 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_maps commit 16f15e5ab2b79590a8ae410f76434aa6690c1fc4
artbio
parents: 16
diff changeset
78 else:
b28dcd4051e8 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_maps commit 16f15e5ab2b79590a8ae410f76434aa6690c1fc4
artbio
parents: 16
diff changeset
79 map_dictionary[(chrom, positions[0]+1, 'F')].append(
b28dcd4051e8 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_maps commit 16f15e5ab2b79590a8ae410f76434aa6690c1fc4
artbio
parents: 16
diff changeset
80 read.query_alignment_length)
0
6d48150495e3 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_maps commit d4d8106d66b65679a1a685ab94bfcf99cdb7b959
artbio
parents:
diff changeset
81 return map_dictionary
6d48150495e3 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_maps commit d4d8106d66b65679a1a685ab94bfcf99cdb7b959
artbio
parents:
diff changeset
82
9
3ea75c573429 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_maps commit 6199193c7fe2cb56403eea8af0b40d44f7311fd5
artbio
parents: 8
diff changeset
83 def grouper(self, iterable, clust_distance):
3ea75c573429 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_maps commit 6199193c7fe2cb56403eea8af0b40d44f7311fd5
artbio
parents: 8
diff changeset
84 prev = None
3ea75c573429 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_maps commit 6199193c7fe2cb56403eea8af0b40d44f7311fd5
artbio
parents: 8
diff changeset
85 group = []
3ea75c573429 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_maps commit 6199193c7fe2cb56403eea8af0b40d44f7311fd5
artbio
parents: 8
diff changeset
86 for item in iterable:
3ea75c573429 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_maps commit 6199193c7fe2cb56403eea8af0b40d44f7311fd5
artbio
parents: 8
diff changeset
87 if not prev or item - prev <= clust_distance:
3ea75c573429 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_maps commit 6199193c7fe2cb56403eea8af0b40d44f7311fd5
artbio
parents: 8
diff changeset
88 group.append(item)
3ea75c573429 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_maps commit 6199193c7fe2cb56403eea8af0b40d44f7311fd5
artbio
parents: 8
diff changeset
89 else:
3ea75c573429 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_maps commit 6199193c7fe2cb56403eea8af0b40d44f7311fd5
artbio
parents: 8
diff changeset
90 yield group
3ea75c573429 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_maps commit 6199193c7fe2cb56403eea8af0b40d44f7311fd5
artbio
parents: 8
diff changeset
91 group = [item]
3ea75c573429 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_maps commit 6199193c7fe2cb56403eea8af0b40d44f7311fd5
artbio
parents: 8
diff changeset
92 prev = item
3ea75c573429 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_maps commit 6199193c7fe2cb56403eea8af0b40d44f7311fd5
artbio
parents: 8
diff changeset
93 if group:
3ea75c573429 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_maps commit 6199193c7fe2cb56403eea8af0b40d44f7311fd5
artbio
parents: 8
diff changeset
94 yield group
3ea75c573429 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_maps commit 6199193c7fe2cb56403eea8af0b40d44f7311fd5
artbio
parents: 8
diff changeset
95
3ea75c573429 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_maps commit 6199193c7fe2cb56403eea8af0b40d44f7311fd5
artbio
parents: 8
diff changeset
96 def tile_map(self, map_dic, clust_distance):
3ea75c573429 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_maps commit 6199193c7fe2cb56403eea8af0b40d44f7311fd5
artbio
parents: 8
diff changeset
97 '''
3ea75c573429 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_maps commit 6199193c7fe2cb56403eea8af0b40d44f7311fd5
artbio
parents: 8
diff changeset
98 takes a map_dictionary {(chromosome,read_position,polarity):
16
600e2498bd21 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_maps commit 82bb0971cde6ba1972588c9315c3007bc3a5a6a7-dirty
artbio
parents: 15
diff changeset
99 [read_length, ...]}
600e2498bd21 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_maps commit 82bb0971cde6ba1972588c9315c3007bc3a5a6a7-dirty
artbio
parents: 15
diff changeset
100 and returns a map_dictionary with structure:
600e2498bd21 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_maps commit 82bb0971cde6ba1972588c9315c3007bc3a5a6a7-dirty
artbio
parents: 15
diff changeset
101 {(chromosome,read_position,polarity):
600e2498bd21 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_maps commit 82bb0971cde6ba1972588c9315c3007bc3a5a6a7-dirty
artbio
parents: 15
diff changeset
102 ([read_length, ...], [start_clust, end_clust])}
9
3ea75c573429 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_maps commit 6199193c7fe2cb56403eea8af0b40d44f7311fd5
artbio
parents: 8
diff changeset
103 '''
3ea75c573429 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_maps commit 6199193c7fe2cb56403eea8af0b40d44f7311fd5
artbio
parents: 8
diff changeset
104 clustered_dic = defaultdict(list)
3ea75c573429 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_maps commit 6199193c7fe2cb56403eea8af0b40d44f7311fd5
artbio
parents: 8
diff changeset
105 for chrom in self.chromosomes:
3ea75c573429 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_maps commit 6199193c7fe2cb56403eea8af0b40d44f7311fd5
artbio
parents: 8
diff changeset
106 F_chrom_coord = []
3ea75c573429 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_maps commit 6199193c7fe2cb56403eea8af0b40d44f7311fd5
artbio
parents: 8
diff changeset
107 R_chrom_coord = []
3ea75c573429 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_maps commit 6199193c7fe2cb56403eea8af0b40d44f7311fd5
artbio
parents: 8
diff changeset
108 for key in map_dic:
3ea75c573429 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_maps commit 6199193c7fe2cb56403eea8af0b40d44f7311fd5
artbio
parents: 8
diff changeset
109 if key[0] == chrom and key[2] == 'F':
3ea75c573429 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_maps commit 6199193c7fe2cb56403eea8af0b40d44f7311fd5
artbio
parents: 8
diff changeset
110 F_chrom_coord.append(key[1])
3ea75c573429 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_maps commit 6199193c7fe2cb56403eea8af0b40d44f7311fd5
artbio
parents: 8
diff changeset
111 elif key[0] == chrom and key[2] == 'R':
3ea75c573429 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_maps commit 6199193c7fe2cb56403eea8af0b40d44f7311fd5
artbio
parents: 8
diff changeset
112 R_chrom_coord.append(key[1])
3ea75c573429 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_maps commit 6199193c7fe2cb56403eea8af0b40d44f7311fd5
artbio
parents: 8
diff changeset
113 F_chrom_coord = list(set(F_chrom_coord))
3ea75c573429 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_maps commit 6199193c7fe2cb56403eea8af0b40d44f7311fd5
artbio
parents: 8
diff changeset
114 R_chrom_coord = list(set(R_chrom_coord))
3ea75c573429 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_maps commit 6199193c7fe2cb56403eea8af0b40d44f7311fd5
artbio
parents: 8
diff changeset
115 F_chrom_coord.sort()
3ea75c573429 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_maps commit 6199193c7fe2cb56403eea8af0b40d44f7311fd5
artbio
parents: 8
diff changeset
116 R_chrom_coord.sort()
3ea75c573429 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_maps commit 6199193c7fe2cb56403eea8af0b40d44f7311fd5
artbio
parents: 8
diff changeset
117 F_clust_values = [i for i in self.grouper(F_chrom_coord,
3ea75c573429 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_maps commit 6199193c7fe2cb56403eea8af0b40d44f7311fd5
artbio
parents: 8
diff changeset
118 clust_distance)]
3ea75c573429 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_maps commit 6199193c7fe2cb56403eea8af0b40d44f7311fd5
artbio
parents: 8
diff changeset
119 F_clust_keys = [(i[-1]+i[0])/2 for i in F_clust_values]
3ea75c573429 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_maps commit 6199193c7fe2cb56403eea8af0b40d44f7311fd5
artbio
parents: 8
diff changeset
120 R_clust_values = [i for i in self.grouper(R_chrom_coord,
3ea75c573429 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_maps commit 6199193c7fe2cb56403eea8af0b40d44f7311fd5
artbio
parents: 8
diff changeset
121 clust_distance)]
3ea75c573429 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_maps commit 6199193c7fe2cb56403eea8af0b40d44f7311fd5
artbio
parents: 8
diff changeset
122 R_clust_keys = [(i[-1]+i[0])/2 for i in R_clust_values]
16
600e2498bd21 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_maps commit 82bb0971cde6ba1972588c9315c3007bc3a5a6a7-dirty
artbio
parents: 15
diff changeset
123 # now 2 dictionnaries (F and R) with structure:
600e2498bd21 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_maps commit 82bb0971cde6ba1972588c9315c3007bc3a5a6a7-dirty
artbio
parents: 15
diff changeset
124 # {centered_coordinate: [coord1, coord2, coord3, ..]}
9
3ea75c573429 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_maps commit 6199193c7fe2cb56403eea8af0b40d44f7311fd5
artbio
parents: 8
diff changeset
125 F_clust_dic = dict(zip(F_clust_keys, F_clust_values))
3ea75c573429 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_maps commit 6199193c7fe2cb56403eea8af0b40d44f7311fd5
artbio
parents: 8
diff changeset
126 R_clust_dic = dict(zip(R_clust_keys, R_clust_values))
3ea75c573429 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_maps commit 6199193c7fe2cb56403eea8af0b40d44f7311fd5
artbio
parents: 8
diff changeset
127 for centcoor in F_clust_dic:
3ea75c573429 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_maps commit 6199193c7fe2cb56403eea8af0b40d44f7311fd5
artbio
parents: 8
diff changeset
128 accumulator = []
3ea75c573429 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_maps commit 6199193c7fe2cb56403eea8af0b40d44f7311fd5
artbio
parents: 8
diff changeset
129 for coor in F_clust_dic[centcoor]:
3ea75c573429 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_maps commit 6199193c7fe2cb56403eea8af0b40d44f7311fd5
artbio
parents: 8
diff changeset
130 accumulator.extend(map_dic[(chrom, coor, 'F')])
16
600e2498bd21 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_maps commit 82bb0971cde6ba1972588c9315c3007bc3a5a6a7-dirty
artbio
parents: 15
diff changeset
131 clustered_dic[(chrom, centcoor, 'F')] = [len(accumulator), [
600e2498bd21 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_maps commit 82bb0971cde6ba1972588c9315c3007bc3a5a6a7-dirty
artbio
parents: 15
diff changeset
132 F_clust_dic[centcoor][0],
600e2498bd21 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_maps commit 82bb0971cde6ba1972588c9315c3007bc3a5a6a7-dirty
artbio
parents: 15
diff changeset
133 F_clust_dic[centcoor][-1]]]
9
3ea75c573429 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_maps commit 6199193c7fe2cb56403eea8af0b40d44f7311fd5
artbio
parents: 8
diff changeset
134 for centcoor in R_clust_dic:
3ea75c573429 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_maps commit 6199193c7fe2cb56403eea8af0b40d44f7311fd5
artbio
parents: 8
diff changeset
135 accumulator = []
3ea75c573429 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_maps commit 6199193c7fe2cb56403eea8af0b40d44f7311fd5
artbio
parents: 8
diff changeset
136 for coor in R_clust_dic[centcoor]:
3ea75c573429 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_maps commit 6199193c7fe2cb56403eea8af0b40d44f7311fd5
artbio
parents: 8
diff changeset
137 accumulator.extend(map_dic[(chrom, coor, 'R')])
16
600e2498bd21 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_maps commit 82bb0971cde6ba1972588c9315c3007bc3a5a6a7-dirty
artbio
parents: 15
diff changeset
138 clustered_dic[(chrom, centcoor, 'R')] = [len(accumulator), [
600e2498bd21 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_maps commit 82bb0971cde6ba1972588c9315c3007bc3a5a6a7-dirty
artbio
parents: 15
diff changeset
139 R_clust_dic[centcoor][0],
600e2498bd21 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_maps commit 82bb0971cde6ba1972588c9315c3007bc3a5a6a7-dirty
artbio
parents: 15
diff changeset
140 R_clust_dic[centcoor][-1]]]
9
3ea75c573429 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_maps commit 6199193c7fe2cb56403eea8af0b40d44f7311fd5
artbio
parents: 8
diff changeset
141 return clustered_dic
3ea75c573429 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_maps commit 6199193c7fe2cb56403eea8af0b40d44f7311fd5
artbio
parents: 8
diff changeset
142
3
ed8b0142538d planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_maps commit 7b2ceb05489c27ddb769c38fdec56274108a6fa1
artbio
parents: 2
diff changeset
143 def compute_readcount(self, map_dictionary, out):
2
507383cce5a8 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_maps commit edbb53cb13b52bf8e71c562fa8acc2c3be2fb270
artbio
parents: 1
diff changeset
144 '''
507383cce5a8 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_maps commit edbb53cb13b52bf8e71c562fa8acc2c3be2fb270
artbio
parents: 1
diff changeset
145 takes a map_dictionary as input and writes
507383cce5a8 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_maps commit edbb53cb13b52bf8e71c562fa8acc2c3be2fb270
artbio
parents: 1
diff changeset
146 a readmap_dictionary {(chromosome,read_position,polarity):
507383cce5a8 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_maps commit edbb53cb13b52bf8e71c562fa8acc2c3be2fb270
artbio
parents: 1
diff changeset
147 number_of_reads}
507383cce5a8 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_maps commit edbb53cb13b52bf8e71c562fa8acc2c3be2fb270
artbio
parents: 1
diff changeset
148 in an open file handler out
0
6d48150495e3 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_maps commit d4d8106d66b65679a1a685ab94bfcf99cdb7b959
artbio
parents:
diff changeset
149 '''
2
507383cce5a8 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_maps commit edbb53cb13b52bf8e71c562fa8acc2c3be2fb270
artbio
parents: 1
diff changeset
150 readmap_dictionary = dict()
507383cce5a8 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_maps commit edbb53cb13b52bf8e71c562fa8acc2c3be2fb270
artbio
parents: 1
diff changeset
151 for key in map_dictionary:
507383cce5a8 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_maps commit edbb53cb13b52bf8e71c562fa8acc2c3be2fb270
artbio
parents: 1
diff changeset
152 readmap_dictionary[key] = len(map_dictionary[key])
507383cce5a8 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_maps commit edbb53cb13b52bf8e71c562fa8acc2c3be2fb270
artbio
parents: 1
diff changeset
153 self.write_table(readmap_dictionary, out)
507383cce5a8 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_maps commit edbb53cb13b52bf8e71c562fa8acc2c3be2fb270
artbio
parents: 1
diff changeset
154
507383cce5a8 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_maps commit edbb53cb13b52bf8e71c562fa8acc2c3be2fb270
artbio
parents: 1
diff changeset
155 def compute_max(self, map_dictionary, out):
507383cce5a8 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_maps commit edbb53cb13b52bf8e71c562fa8acc2c3be2fb270
artbio
parents: 1
diff changeset
156 '''
507383cce5a8 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_maps commit edbb53cb13b52bf8e71c562fa8acc2c3be2fb270
artbio
parents: 1
diff changeset
157 takes a map_dictionary as input and writes
0
6d48150495e3 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_maps commit d4d8106d66b65679a1a685ab94bfcf99cdb7b959
artbio
parents:
diff changeset
158 a max_dictionary {(chromosome,read_position,polarity):
6d48150495e3 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_maps commit d4d8106d66b65679a1a685ab94bfcf99cdb7b959
artbio
parents:
diff changeset
159 max_of_number_of_read_at_any_position}
2
507383cce5a8 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_maps commit edbb53cb13b52bf8e71c562fa8acc2c3be2fb270
artbio
parents: 1
diff changeset
160 Not clear this function is still required
0
6d48150495e3 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_maps commit d4d8106d66b65679a1a685ab94bfcf99cdb7b959
artbio
parents:
diff changeset
161 '''
6d48150495e3 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_maps commit d4d8106d66b65679a1a685ab94bfcf99cdb7b959
artbio
parents:
diff changeset
162 merge_keylist = [(i[0], 0) for i in map_dictionary.keys()]
6d48150495e3 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_maps commit d4d8106d66b65679a1a685ab94bfcf99cdb7b959
artbio
parents:
diff changeset
163 max_dictionary = dict(merge_keylist)
6d48150495e3 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_maps commit d4d8106d66b65679a1a685ab94bfcf99cdb7b959
artbio
parents:
diff changeset
164 for key in map_dictionary:
6d48150495e3 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_maps commit d4d8106d66b65679a1a685ab94bfcf99cdb7b959
artbio
parents:
diff changeset
165 if len(map_dictionary[key]) > max_dictionary[key[0]]:
6d48150495e3 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_maps commit d4d8106d66b65679a1a685ab94bfcf99cdb7b959
artbio
parents:
diff changeset
166 max_dictionary[key[0]] = len(map_dictionary[key])
2
507383cce5a8 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_maps commit edbb53cb13b52bf8e71c562fa8acc2c3be2fb270
artbio
parents: 1
diff changeset
167 self.write_table(max_dictionary, out)
0
6d48150495e3 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_maps commit d4d8106d66b65679a1a685ab94bfcf99cdb7b959
artbio
parents:
diff changeset
168
2
507383cce5a8 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_maps commit edbb53cb13b52bf8e71c562fa8acc2c3be2fb270
artbio
parents: 1
diff changeset
169 def compute_mean(self, map_dictionary, out):
0
6d48150495e3 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_maps commit d4d8106d66b65679a1a685ab94bfcf99cdb7b959
artbio
parents:
diff changeset
170 '''
6d48150495e3 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_maps commit d4d8106d66b65679a1a685ab94bfcf99cdb7b959
artbio
parents:
diff changeset
171 takes a map_dictionary as input and returns
6d48150495e3 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_maps commit d4d8106d66b65679a1a685ab94bfcf99cdb7b959
artbio
parents:
diff changeset
172 a mean_dictionary {(chromosome,read_position,polarity):
6d48150495e3 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_maps commit d4d8106d66b65679a1a685ab94bfcf99cdb7b959
artbio
parents:
diff changeset
173 mean_value_of_reads}
6d48150495e3 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_maps commit d4d8106d66b65679a1a685ab94bfcf99cdb7b959
artbio
parents:
diff changeset
174 '''
6d48150495e3 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_maps commit d4d8106d66b65679a1a685ab94bfcf99cdb7b959
artbio
parents:
diff changeset
175 mean_dictionary = dict()
6d48150495e3 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_maps commit d4d8106d66b65679a1a685ab94bfcf99cdb7b959
artbio
parents:
diff changeset
176 for key in map_dictionary:
6d48150495e3 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_maps commit d4d8106d66b65679a1a685ab94bfcf99cdb7b959
artbio
parents:
diff changeset
177 if len(map_dictionary[key]) == 0:
6d48150495e3 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_maps commit d4d8106d66b65679a1a685ab94bfcf99cdb7b959
artbio
parents:
diff changeset
178 mean_dictionary[key] = 0
6d48150495e3 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_maps commit d4d8106d66b65679a1a685ab94bfcf99cdb7b959
artbio
parents:
diff changeset
179 else:
6d48150495e3 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_maps commit d4d8106d66b65679a1a685ab94bfcf99cdb7b959
artbio
parents:
diff changeset
180 mean_dictionary[key] = round(numpy.mean(map_dictionary[key]),
6d48150495e3 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_maps commit d4d8106d66b65679a1a685ab94bfcf99cdb7b959
artbio
parents:
diff changeset
181 1)
2
507383cce5a8 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_maps commit edbb53cb13b52bf8e71c562fa8acc2c3be2fb270
artbio
parents: 1
diff changeset
182 self.write_table(mean_dictionary, out)
0
6d48150495e3 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_maps commit d4d8106d66b65679a1a685ab94bfcf99cdb7b959
artbio
parents:
diff changeset
183
2
507383cce5a8 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_maps commit edbb53cb13b52bf8e71c562fa8acc2c3be2fb270
artbio
parents: 1
diff changeset
184 def compute_median(self, map_dictionary, out):
0
6d48150495e3 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_maps commit d4d8106d66b65679a1a685ab94bfcf99cdb7b959
artbio
parents:
diff changeset
185 '''
6d48150495e3 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_maps commit d4d8106d66b65679a1a685ab94bfcf99cdb7b959
artbio
parents:
diff changeset
186 takes a map_dictionary as input and returns
6d48150495e3 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_maps commit d4d8106d66b65679a1a685ab94bfcf99cdb7b959
artbio
parents:
diff changeset
187 a mean_dictionary {(chromosome,read_position,polarity):
6d48150495e3 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_maps commit d4d8106d66b65679a1a685ab94bfcf99cdb7b959
artbio
parents:
diff changeset
188 mean_value_of_reads}
6d48150495e3 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_maps commit d4d8106d66b65679a1a685ab94bfcf99cdb7b959
artbio
parents:
diff changeset
189 '''
6d48150495e3 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_maps commit d4d8106d66b65679a1a685ab94bfcf99cdb7b959
artbio
parents:
diff changeset
190 median_dictionary = dict()
6d48150495e3 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_maps commit d4d8106d66b65679a1a685ab94bfcf99cdb7b959
artbio
parents:
diff changeset
191 for key in map_dictionary:
6d48150495e3 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_maps commit d4d8106d66b65679a1a685ab94bfcf99cdb7b959
artbio
parents:
diff changeset
192 if len(map_dictionary[key]) == 0:
6d48150495e3 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_maps commit d4d8106d66b65679a1a685ab94bfcf99cdb7b959
artbio
parents:
diff changeset
193 median_dictionary[key] = 0
6d48150495e3 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_maps commit d4d8106d66b65679a1a685ab94bfcf99cdb7b959
artbio
parents:
diff changeset
194 else:
6d48150495e3 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_maps commit d4d8106d66b65679a1a685ab94bfcf99cdb7b959
artbio
parents:
diff changeset
195 median_dictionary[key] = numpy.median(map_dictionary[key])
2
507383cce5a8 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_maps commit edbb53cb13b52bf8e71c562fa8acc2c3be2fb270
artbio
parents: 1
diff changeset
196 self.write_table(median_dictionary, out)
0
6d48150495e3 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_maps commit d4d8106d66b65679a1a685ab94bfcf99cdb7b959
artbio
parents:
diff changeset
197
16
600e2498bd21 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_maps commit 82bb0971cde6ba1972588c9315c3007bc3a5a6a7-dirty
artbio
parents: 15
diff changeset
198 def compute_coverage(self, map_dictionary, out, quality=15):
0
6d48150495e3 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_maps commit d4d8106d66b65679a1a685ab94bfcf99cdb7b959
artbio
parents:
diff changeset
199 '''
6d48150495e3 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_maps commit d4d8106d66b65679a1a685ab94bfcf99cdb7b959
artbio
parents:
diff changeset
200 takes a map_dictionary as input and returns
6d48150495e3 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_maps commit d4d8106d66b65679a1a685ab94bfcf99cdb7b959
artbio
parents:
diff changeset
201 a coverage_dictionary {(chromosome,read_position,polarity):
6d48150495e3 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_maps commit d4d8106d66b65679a1a685ab94bfcf99cdb7b959
artbio
parents:
diff changeset
202 coverage}
6d48150495e3 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_maps commit d4d8106d66b65679a1a685ab94bfcf99cdb7b959
artbio
parents:
diff changeset
203 '''
6d48150495e3 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_maps commit d4d8106d66b65679a1a685ab94bfcf99cdb7b959
artbio
parents:
diff changeset
204 coverage_dictionary = dict()
6d48150495e3 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_maps commit d4d8106d66b65679a1a685ab94bfcf99cdb7b959
artbio
parents:
diff changeset
205 for chrom in self.chromosomes:
6d48150495e3 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_maps commit d4d8106d66b65679a1a685ab94bfcf99cdb7b959
artbio
parents:
diff changeset
206 coverage_dictionary[(chrom, 1, 'F')] = 0
6d48150495e3 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_maps commit d4d8106d66b65679a1a685ab94bfcf99cdb7b959
artbio
parents:
diff changeset
207 coverage_dictionary[(chrom, self.chromosomes[chrom], 'F')] = 0
4
a6b9a081064b planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_maps commit b58cb36616bf351278e57ec1949e9ebf3c3cdff1
artbio
parents: 3
diff changeset
208 for read in self.bam_object.fetch(chrom):
a6b9a081064b planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_maps commit b58cb36616bf351278e57ec1949e9ebf3c3cdff1
artbio
parents: 3
diff changeset
209 positions = read.positions # a list of covered positions
a6b9a081064b planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_maps commit b58cb36616bf351278e57ec1949e9ebf3c3cdff1
artbio
parents: 3
diff changeset
210 for pos in positions:
a6b9a081064b planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_maps commit b58cb36616bf351278e57ec1949e9ebf3c3cdff1
artbio
parents: 3
diff changeset
211 if not map_dictionary[(chrom, pos+1, 'F')]:
a6b9a081064b planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_maps commit b58cb36616bf351278e57ec1949e9ebf3c3cdff1
artbio
parents: 3
diff changeset
212 map_dictionary[(chrom, pos+1, 'F')] = []
0
6d48150495e3 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_maps commit d4d8106d66b65679a1a685ab94bfcf99cdb7b959
artbio
parents:
diff changeset
213 for key in map_dictionary:
6d48150495e3 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_maps commit d4d8106d66b65679a1a685ab94bfcf99cdb7b959
artbio
parents:
diff changeset
214 coverage = self.bam_object.count_coverage(
6d48150495e3 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_maps commit d4d8106d66b65679a1a685ab94bfcf99cdb7b959
artbio
parents:
diff changeset
215 reference=key[0],
6d48150495e3 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_maps commit d4d8106d66b65679a1a685ab94bfcf99cdb7b959
artbio
parents:
diff changeset
216 start=key[1]-1,
6d48150495e3 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_maps commit d4d8106d66b65679a1a685ab94bfcf99cdb7b959
artbio
parents:
diff changeset
217 end=key[1],
6d48150495e3 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_maps commit d4d8106d66b65679a1a685ab94bfcf99cdb7b959
artbio
parents:
diff changeset
218 quality_threshold=quality)
6d48150495e3 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_maps commit d4d8106d66b65679a1a685ab94bfcf99cdb7b959
artbio
parents:
diff changeset
219 """ Add the 4 coverage values """
6d48150495e3 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_maps commit d4d8106d66b65679a1a685ab94bfcf99cdb7b959
artbio
parents:
diff changeset
220 coverage = [sum(x) for x in zip(*coverage)]
6d48150495e3 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_maps commit d4d8106d66b65679a1a685ab94bfcf99cdb7b959
artbio
parents:
diff changeset
221 coverage_dictionary[key] = coverage[0]
2
507383cce5a8 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_maps commit edbb53cb13b52bf8e71c562fa8acc2c3be2fb270
artbio
parents: 1
diff changeset
222 self.write_table(coverage_dictionary, out)
0
6d48150495e3 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_maps commit d4d8106d66b65679a1a685ab94bfcf99cdb7b959
artbio
parents:
diff changeset
223
2
507383cce5a8 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_maps commit edbb53cb13b52bf8e71c562fa8acc2c3be2fb270
artbio
parents: 1
diff changeset
224 def compute_size(self, map_dictionary, out):
0
6d48150495e3 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_maps commit d4d8106d66b65679a1a685ab94bfcf99cdb7b959
artbio
parents:
diff changeset
225 '''
6d48150495e3 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_maps commit d4d8106d66b65679a1a685ab94bfcf99cdb7b959
artbio
parents:
diff changeset
226 Takes a map_dictionary and returns a dictionary of sizes:
6d48150495e3 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_maps commit d4d8106d66b65679a1a685ab94bfcf99cdb7b959
artbio
parents:
diff changeset
227 {chrom: {polarity: {size: nbre of reads}}}
6d48150495e3 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_maps commit d4d8106d66b65679a1a685ab94bfcf99cdb7b959
artbio
parents:
diff changeset
228 '''
6d48150495e3 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_maps commit d4d8106d66b65679a1a685ab94bfcf99cdb7b959
artbio
parents:
diff changeset
229 size_dictionary = defaultdict(lambda: defaultdict(
1
40972a8dfab9 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_maps commit ad60e6655aabe30246043d95f14646b2527c9255
artbio
parents: 0
diff changeset
230 lambda: defaultdict(int)))
0
6d48150495e3 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_maps commit d4d8106d66b65679a1a685ab94bfcf99cdb7b959
artbio
parents:
diff changeset
231 # to track empty chromosomes
6d48150495e3 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_maps commit d4d8106d66b65679a1a685ab94bfcf99cdb7b959
artbio
parents:
diff changeset
232 for chrom in self.chromosomes:
6d48150495e3 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_maps commit d4d8106d66b65679a1a685ab94bfcf99cdb7b959
artbio
parents:
diff changeset
233 if self.bam_object.count(chrom) == 0:
6d48150495e3 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_maps commit d4d8106d66b65679a1a685ab94bfcf99cdb7b959
artbio
parents:
diff changeset
234 size_dictionary[chrom]['F'][10] = 0
6d48150495e3 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_maps commit d4d8106d66b65679a1a685ab94bfcf99cdb7b959
artbio
parents:
diff changeset
235 for key in map_dictionary:
6d48150495e3 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_maps commit d4d8106d66b65679a1a685ab94bfcf99cdb7b959
artbio
parents:
diff changeset
236 for size in map_dictionary[key]:
6d48150495e3 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_maps commit d4d8106d66b65679a1a685ab94bfcf99cdb7b959
artbio
parents:
diff changeset
237 size_dictionary[key[0]][key[2]][size] += 1
2
507383cce5a8 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_maps commit edbb53cb13b52bf8e71c562fa8acc2c3be2fb270
artbio
parents: 1
diff changeset
238 self.write_size_table(size_dictionary, out)
0
6d48150495e3 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_maps commit d4d8106d66b65679a1a685ab94bfcf99cdb7b959
artbio
parents:
diff changeset
239
2
507383cce5a8 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_maps commit edbb53cb13b52bf8e71c562fa8acc2c3be2fb270
artbio
parents: 1
diff changeset
240 def write_table(self, mapdict, out):
0
6d48150495e3 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_maps commit d4d8106d66b65679a1a685ab94bfcf99cdb7b959
artbio
parents:
diff changeset
241 '''
16
600e2498bd21 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_maps commit 82bb0971cde6ba1972588c9315c3007bc3a5a6a7-dirty
artbio
parents: 15
diff changeset
242 Writer of a tabular file
2
507383cce5a8 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_maps commit edbb53cb13b52bf8e71c562fa8acc2c3be2fb270
artbio
parents: 1
diff changeset
243 Dataset, Chromosome, Chrom_length, Coordinate, Polarity,
507383cce5a8 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_maps commit edbb53cb13b52bf8e71c562fa8acc2c3be2fb270
artbio
parents: 1
diff changeset
244 <some mapped value>
0
6d48150495e3 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_maps commit d4d8106d66b65679a1a685ab94bfcf99cdb7b959
artbio
parents:
diff changeset
245 out is an *open* file handler
6d48150495e3 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_maps commit d4d8106d66b65679a1a685ab94bfcf99cdb7b959
artbio
parents:
diff changeset
246 '''
2
507383cce5a8 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_maps commit edbb53cb13b52bf8e71c562fa8acc2c3be2fb270
artbio
parents: 1
diff changeset
247 for key in sorted(mapdict):
507383cce5a8 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_maps commit edbb53cb13b52bf8e71c562fa8acc2c3be2fb270
artbio
parents: 1
diff changeset
248 line = [self.sample_name, key[0], self.chromosomes[key[0]],
507383cce5a8 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_maps commit edbb53cb13b52bf8e71c562fa8acc2c3be2fb270
artbio
parents: 1
diff changeset
249 key[1], key[2], mapdict[key]]
507383cce5a8 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_maps commit edbb53cb13b52bf8e71c562fa8acc2c3be2fb270
artbio
parents: 1
diff changeset
250 line = [str(i) for i in line]
507383cce5a8 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_maps commit edbb53cb13b52bf8e71c562fa8acc2c3be2fb270
artbio
parents: 1
diff changeset
251 out.write('\t'.join(line) + '\n')
507383cce5a8 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_maps commit edbb53cb13b52bf8e71c562fa8acc2c3be2fb270
artbio
parents: 1
diff changeset
252
507383cce5a8 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_maps commit edbb53cb13b52bf8e71c562fa8acc2c3be2fb270
artbio
parents: 1
diff changeset
253 def write_size_table(self, sizedic, out):
507383cce5a8 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_maps commit edbb53cb13b52bf8e71c562fa8acc2c3be2fb270
artbio
parents: 1
diff changeset
254 '''
16
600e2498bd21 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_maps commit 82bb0971cde6ba1972588c9315c3007bc3a5a6a7-dirty
artbio
parents: 15
diff changeset
255 Writer of a tabular file
600e2498bd21 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_maps commit 82bb0971cde6ba1972588c9315c3007bc3a5a6a7-dirty
artbio
parents: 15
diff changeset
256 Dataset, Chromosome, Chrom_length, <category (size)>, <some value>
2
507383cce5a8 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_maps commit edbb53cb13b52bf8e71c562fa8acc2c3be2fb270
artbio
parents: 1
diff changeset
257 out is an *open* file handler
507383cce5a8 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_maps commit edbb53cb13b52bf8e71c562fa8acc2c3be2fb270
artbio
parents: 1
diff changeset
258 '''
507383cce5a8 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_maps commit edbb53cb13b52bf8e71c562fa8acc2c3be2fb270
artbio
parents: 1
diff changeset
259 for chrom in sorted(sizedic):
507383cce5a8 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_maps commit edbb53cb13b52bf8e71c562fa8acc2c3be2fb270
artbio
parents: 1
diff changeset
260 sizes = sizedic[chrom]['F'].keys()
507383cce5a8 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_maps commit edbb53cb13b52bf8e71c562fa8acc2c3be2fb270
artbio
parents: 1
diff changeset
261 sizes.extend(sizedic[chrom]['R'].keys())
507383cce5a8 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_maps commit edbb53cb13b52bf8e71c562fa8acc2c3be2fb270
artbio
parents: 1
diff changeset
262 for polarity in sorted(sizedic[chrom]):
0
6d48150495e3 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_maps commit d4d8106d66b65679a1a685ab94bfcf99cdb7b959
artbio
parents:
diff changeset
263 for size in range(min(sizes), max(sizes)+1):
6d48150495e3 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_maps commit d4d8106d66b65679a1a685ab94bfcf99cdb7b959
artbio
parents:
diff changeset
264 try:
6d48150495e3 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_maps commit d4d8106d66b65679a1a685ab94bfcf99cdb7b959
artbio
parents:
diff changeset
265 line = [self.sample_name, chrom, polarity, size,
2
507383cce5a8 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_maps commit edbb53cb13b52bf8e71c562fa8acc2c3be2fb270
artbio
parents: 1
diff changeset
266 sizedic[chrom][polarity][size]]
0
6d48150495e3 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_maps commit d4d8106d66b65679a1a685ab94bfcf99cdb7b959
artbio
parents:
diff changeset
267 except KeyError:
6d48150495e3 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_maps commit d4d8106d66b65679a1a685ab94bfcf99cdb7b959
artbio
parents:
diff changeset
268 line = [self.sample_name, chrom, polarity, size, 0]
6d48150495e3 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_maps commit d4d8106d66b65679a1a685ab94bfcf99cdb7b959
artbio
parents:
diff changeset
269 line = [str(i) for i in line]
6d48150495e3 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_maps commit d4d8106d66b65679a1a685ab94bfcf99cdb7b959
artbio
parents:
diff changeset
270 out.write('\t'.join(line) + '\n')
6d48150495e3 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_maps commit d4d8106d66b65679a1a685ab94bfcf99cdb7b959
artbio
parents:
diff changeset
271
16
600e2498bd21 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_maps commit 82bb0971cde6ba1972588c9315c3007bc3a5a6a7-dirty
artbio
parents: 15
diff changeset
272 def write_cluster_table(self, clustered_dic, out):
600e2498bd21 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_maps commit 82bb0971cde6ba1972588c9315c3007bc3a5a6a7-dirty
artbio
parents: 15
diff changeset
273 '''
600e2498bd21 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_maps commit 82bb0971cde6ba1972588c9315c3007bc3a5a6a7-dirty
artbio
parents: 15
diff changeset
274 Writer of a tabular file
600e2498bd21 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_maps commit 82bb0971cde6ba1972588c9315c3007bc3a5a6a7-dirty
artbio
parents: 15
diff changeset
275 Dataset, Chromosome, Chrom_length, Coordinate, Polarity,
600e2498bd21 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_maps commit 82bb0971cde6ba1972588c9315c3007bc3a5a6a7-dirty
artbio
parents: 15
diff changeset
276 <some mapped value>
600e2498bd21 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_maps commit 82bb0971cde6ba1972588c9315c3007bc3a5a6a7-dirty
artbio
parents: 15
diff changeset
277 out is an *open* file handler
600e2498bd21 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_maps commit 82bb0971cde6ba1972588c9315c3007bc3a5a6a7-dirty
artbio
parents: 15
diff changeset
278 '''
600e2498bd21 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_maps commit 82bb0971cde6ba1972588c9315c3007bc3a5a6a7-dirty
artbio
parents: 15
diff changeset
279 for key in sorted(clustered_dic):
600e2498bd21 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_maps commit 82bb0971cde6ba1972588c9315c3007bc3a5a6a7-dirty
artbio
parents: 15
diff changeset
280 start = clustered_dic[key][1][0]
600e2498bd21 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_maps commit 82bb0971cde6ba1972588c9315c3007bc3a5a6a7-dirty
artbio
parents: 15
diff changeset
281 end = clustered_dic[key][1][1]
600e2498bd21 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_maps commit 82bb0971cde6ba1972588c9315c3007bc3a5a6a7-dirty
artbio
parents: 15
diff changeset
282 size = end - start + 1
600e2498bd21 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_maps commit 82bb0971cde6ba1972588c9315c3007bc3a5a6a7-dirty
artbio
parents: 15
diff changeset
283 density = float(clustered_dic[key][0]) / size
600e2498bd21 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_maps commit 82bb0971cde6ba1972588c9315c3007bc3a5a6a7-dirty
artbio
parents: 15
diff changeset
284 line = [self.sample_name, key[0], self.chromosomes[key[0]],
600e2498bd21 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_maps commit 82bb0971cde6ba1972588c9315c3007bc3a5a6a7-dirty
artbio
parents: 15
diff changeset
285 key[1], key[2], clustered_dic[key][0],
600e2498bd21 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_maps commit 82bb0971cde6ba1972588c9315c3007bc3a5a6a7-dirty
artbio
parents: 15
diff changeset
286 str(start) + "-" + str(end), str(size), str(density)]
600e2498bd21 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_maps commit 82bb0971cde6ba1972588c9315c3007bc3a5a6a7-dirty
artbio
parents: 15
diff changeset
287 line = [str(i) for i in line]
600e2498bd21 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_maps commit 82bb0971cde6ba1972588c9315c3007bc3a5a6a7-dirty
artbio
parents: 15
diff changeset
288 out.write('\t'.join(line) + '\n')
600e2498bd21 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_maps commit 82bb0971cde6ba1972588c9315c3007bc3a5a6a7-dirty
artbio
parents: 15
diff changeset
289
0
6d48150495e3 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_maps commit d4d8106d66b65679a1a685ab94bfcf99cdb7b959
artbio
parents:
diff changeset
290
17
b28dcd4051e8 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_maps commit 16f15e5ab2b79590a8ae410f76434aa6690c1fc4
artbio
parents: 16
diff changeset
291 def main(inputs, samples, methods, outputs, minsize, maxsize, cluster,
b28dcd4051e8 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_maps commit 16f15e5ab2b79590a8ae410f76434aa6690c1fc4
artbio
parents: 16
diff changeset
292 nostrand):
2
507383cce5a8 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_maps commit edbb53cb13b52bf8e71c562fa8acc2c3be2fb270
artbio
parents: 1
diff changeset
293 for method, output in zip(methods, outputs):
16
600e2498bd21 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_maps commit 82bb0971cde6ba1972588c9315c3007bc3a5a6a7-dirty
artbio
parents: 15
diff changeset
294 out = open(output, 'w')
2
507383cce5a8 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_maps commit edbb53cb13b52bf8e71c562fa8acc2c3be2fb270
artbio
parents: 1
diff changeset
295 if method == 'Size':
5
12c14642e6ac planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_maps commit 24a21619d79d83b38cef7f1a7b858c621e4c8449
artbio
parents: 4
diff changeset
296 header = ["Dataset", "Chromosome", "Polarity", method, "Counts"]
16
600e2498bd21 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_maps commit 82bb0971cde6ba1972588c9315c3007bc3a5a6a7-dirty
artbio
parents: 15
diff changeset
297 elif cluster:
600e2498bd21 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_maps commit 82bb0971cde6ba1972588c9315c3007bc3a5a6a7-dirty
artbio
parents: 15
diff changeset
298 header = ["Dataset", "Chromosome", "Chrom_length", "Coordinate",
600e2498bd21 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_maps commit 82bb0971cde6ba1972588c9315c3007bc3a5a6a7-dirty
artbio
parents: 15
diff changeset
299 "Polarity", method, "Start-End", "Cluster Size",
600e2498bd21 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_maps commit 82bb0971cde6ba1972588c9315c3007bc3a5a6a7-dirty
artbio
parents: 15
diff changeset
300 "density"]
2
507383cce5a8 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_maps commit edbb53cb13b52bf8e71c562fa8acc2c3be2fb270
artbio
parents: 1
diff changeset
301 else:
507383cce5a8 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_maps commit edbb53cb13b52bf8e71c562fa8acc2c3be2fb270
artbio
parents: 1
diff changeset
302 header = ["Dataset", "Chromosome", "Chrom_length", "Coordinate",
507383cce5a8 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_maps commit edbb53cb13b52bf8e71c562fa8acc2c3be2fb270
artbio
parents: 1
diff changeset
303 "Polarity", method]
16
600e2498bd21 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_maps commit 82bb0971cde6ba1972588c9315c3007bc3a5a6a7-dirty
artbio
parents: 15
diff changeset
304 out.write('\t'.join(header) + '\n')
2
507383cce5a8 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_maps commit edbb53cb13b52bf8e71c562fa8acc2c3be2fb270
artbio
parents: 1
diff changeset
305 for input, sample in zip(inputs, samples):
17
b28dcd4051e8 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_maps commit 16f15e5ab2b79590a8ae410f76434aa6690c1fc4
artbio
parents: 16
diff changeset
306 mapobj = Map(input, sample, minsize, maxsize, cluster, nostrand)
3
ed8b0142538d planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_maps commit 7b2ceb05489c27ddb769c38fdec56274108a6fa1
artbio
parents: 2
diff changeset
307 token = {"Counts": mapobj.compute_readcount,
2
507383cce5a8 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_maps commit edbb53cb13b52bf8e71c562fa8acc2c3be2fb270
artbio
parents: 1
diff changeset
308 "Max": mapobj.compute_max,
507383cce5a8 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_maps commit edbb53cb13b52bf8e71c562fa8acc2c3be2fb270
artbio
parents: 1
diff changeset
309 "Mean": mapobj.compute_mean,
507383cce5a8 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_maps commit edbb53cb13b52bf8e71c562fa8acc2c3be2fb270
artbio
parents: 1
diff changeset
310 "Median": mapobj.compute_median,
507383cce5a8 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_maps commit edbb53cb13b52bf8e71c562fa8acc2c3be2fb270
artbio
parents: 1
diff changeset
311 "Coverage": mapobj.compute_coverage,
16
600e2498bd21 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_maps commit 82bb0971cde6ba1972588c9315c3007bc3a5a6a7-dirty
artbio
parents: 15
diff changeset
312 "Size": mapobj.compute_size,
600e2498bd21 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_maps commit 82bb0971cde6ba1972588c9315c3007bc3a5a6a7-dirty
artbio
parents: 15
diff changeset
313 "cluster": mapobj.write_cluster_table}
600e2498bd21 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_maps commit 82bb0971cde6ba1972588c9315c3007bc3a5a6a7-dirty
artbio
parents: 15
diff changeset
314 if cluster:
600e2498bd21 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_maps commit 82bb0971cde6ba1972588c9315c3007bc3a5a6a7-dirty
artbio
parents: 15
diff changeset
315 token["cluster"](mapobj.map_dict, out)
600e2498bd21 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_maps commit 82bb0971cde6ba1972588c9315c3007bc3a5a6a7-dirty
artbio
parents: 15
diff changeset
316 else:
600e2498bd21 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_maps commit 82bb0971cde6ba1972588c9315c3007bc3a5a6a7-dirty
artbio
parents: 15
diff changeset
317 token[method](mapobj.map_dict, out)
600e2498bd21 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_maps commit 82bb0971cde6ba1972588c9315c3007bc3a5a6a7-dirty
artbio
parents: 15
diff changeset
318 # mapobj.compute_coverage(mapobj.map_dict, out)
600e2498bd21 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_maps commit 82bb0971cde6ba1972588c9315c3007bc3a5a6a7-dirty
artbio
parents: 15
diff changeset
319 out.close()
0
6d48150495e3 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_maps commit d4d8106d66b65679a1a685ab94bfcf99cdb7b959
artbio
parents:
diff changeset
320
6d48150495e3 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_maps commit d4d8106d66b65679a1a685ab94bfcf99cdb7b959
artbio
parents:
diff changeset
321
6d48150495e3 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_maps commit d4d8106d66b65679a1a685ab94bfcf99cdb7b959
artbio
parents:
diff changeset
322 if __name__ == "__main__":
6d48150495e3 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_maps commit d4d8106d66b65679a1a685ab94bfcf99cdb7b959
artbio
parents:
diff changeset
323 args = Parser()
6d48150495e3 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_maps commit d4d8106d66b65679a1a685ab94bfcf99cdb7b959
artbio
parents:
diff changeset
324 # if identical sample names
2
507383cce5a8 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_maps commit edbb53cb13b52bf8e71c562fa8acc2c3be2fb270
artbio
parents: 1
diff changeset
325 if len(set(args.sample_names)) != len(args.sample_names):
507383cce5a8 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_maps commit edbb53cb13b52bf8e71c562fa8acc2c3be2fb270
artbio
parents: 1
diff changeset
326 args.sample_names = [name + '_' + str(i) for
507383cce5a8 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_maps commit edbb53cb13b52bf8e71c562fa8acc2c3be2fb270
artbio
parents: 1
diff changeset
327 i, name in enumerate(args.sample_names)]
8
1827b74f872b planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_maps commit e4588eb6c329e4516e9bcfa084a383be81b55c60
artbio
parents: 5
diff changeset
328 main(args.inputs, args.sample_names, args.plot_methods, args.outputs,
17
b28dcd4051e8 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/small_rna_maps commit 16f15e5ab2b79590a8ae410f76434aa6690c1fc4
artbio
parents: 16
diff changeset
329 args.minsize, args.maxsize, args.cluster, args.nostrand)