annotate mircounts.py @ 2:f59c643b00fc draft

planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/mircounts commit 931c1ee22cd5b8614156894a9bc0ef7ab90e816e
author artbio
date Tue, 25 Jul 2017 11:04:28 -0400
parents da29af78a960
children da1aa7de2b19
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
rev   line source
0
da29af78a960 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/mircounts commit d4d8106d66b65679a1a685ab94bfcf99cdb7b959
artbio
parents:
diff changeset
1 #!/usr/bin/env python
da29af78a960 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/mircounts commit d4d8106d66b65679a1a685ab94bfcf99cdb7b959
artbio
parents:
diff changeset
2 import argparse
da29af78a960 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/mircounts commit d4d8106d66b65679a1a685ab94bfcf99cdb7b959
artbio
parents:
diff changeset
3
da29af78a960 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/mircounts commit d4d8106d66b65679a1a685ab94bfcf99cdb7b959
artbio
parents:
diff changeset
4 import pysam
da29af78a960 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/mircounts commit d4d8106d66b65679a1a685ab94bfcf99cdb7b959
artbio
parents:
diff changeset
5
da29af78a960 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/mircounts commit d4d8106d66b65679a1a685ab94bfcf99cdb7b959
artbio
parents:
diff changeset
6
da29af78a960 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/mircounts commit d4d8106d66b65679a1a685ab94bfcf99cdb7b959
artbio
parents:
diff changeset
7 def Parser():
da29af78a960 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/mircounts commit d4d8106d66b65679a1a685ab94bfcf99cdb7b959
artbio
parents:
diff changeset
8 parser = argparse.ArgumentParser(description='miRNAs counts and coverages')
da29af78a960 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/mircounts commit d4d8106d66b65679a1a685ab94bfcf99cdb7b959
artbio
parents:
diff changeset
9 parser.add_argument('-a', '--alignment', metavar='FILE', type=str,
da29af78a960 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/mircounts commit d4d8106d66b65679a1a685ab94bfcf99cdb7b959
artbio
parents:
diff changeset
10 dest='alignment_file', help='Alignment bam file')
da29af78a960 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/mircounts commit d4d8106d66b65679a1a685ab94bfcf99cdb7b959
artbio
parents:
diff changeset
11 parser.add_argument('--gff', metavar='FILE', type=str, dest='gff_file',
da29af78a960 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/mircounts commit d4d8106d66b65679a1a685ab94bfcf99cdb7b959
artbio
parents:
diff changeset
12 help='GFF3 describing both pre-miRNAs\
da29af78a960 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/mircounts commit d4d8106d66b65679a1a685ab94bfcf99cdb7b959
artbio
parents:
diff changeset
13 and mature miRNAs')
da29af78a960 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/mircounts commit d4d8106d66b65679a1a685ab94bfcf99cdb7b959
artbio
parents:
diff changeset
14 parser.add_argument('-q', '--quality_threshold', type=int,
da29af78a960 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/mircounts commit d4d8106d66b65679a1a685ab94bfcf99cdb7b959
artbio
parents:
diff changeset
15 dest='quality_threshold',
da29af78a960 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/mircounts commit d4d8106d66b65679a1a685ab94bfcf99cdb7b959
artbio
parents:
diff changeset
16 help='Quality threshold for coverage (default=10)',
da29af78a960 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/mircounts commit d4d8106d66b65679a1a685ab94bfcf99cdb7b959
artbio
parents:
diff changeset
17 default=10)
da29af78a960 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/mircounts commit d4d8106d66b65679a1a685ab94bfcf99cdb7b959
artbio
parents:
diff changeset
18 parser.add_argument('-p', '--pre_mirs', type=str, dest='pre_mirs',
da29af78a960 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/mircounts commit d4d8106d66b65679a1a685ab94bfcf99cdb7b959
artbio
parents:
diff changeset
19 help='pre-miRNAs count file path', metavar='FILE')
da29af78a960 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/mircounts commit d4d8106d66b65679a1a685ab94bfcf99cdb7b959
artbio
parents:
diff changeset
20 parser.add_argument('-m', '--mirs', type=str, dest='mirs',
da29af78a960 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/mircounts commit d4d8106d66b65679a1a685ab94bfcf99cdb7b959
artbio
parents:
diff changeset
21 help='mature miRNA count file path', metavar='FILE')
da29af78a960 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/mircounts commit d4d8106d66b65679a1a685ab94bfcf99cdb7b959
artbio
parents:
diff changeset
22 parser.add_argument('--lattice', metavar='FILE', type=str, dest='lattice',
da29af78a960 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/mircounts commit d4d8106d66b65679a1a685ab94bfcf99cdb7b959
artbio
parents:
diff changeset
23 help='Output file for the lattice dataframe.')
da29af78a960 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/mircounts commit d4d8106d66b65679a1a685ab94bfcf99cdb7b959
artbio
parents:
diff changeset
24 args = parser.parse_args()
da29af78a960 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/mircounts commit d4d8106d66b65679a1a685ab94bfcf99cdb7b959
artbio
parents:
diff changeset
25 return args
da29af78a960 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/mircounts commit d4d8106d66b65679a1a685ab94bfcf99cdb7b959
artbio
parents:
diff changeset
26
da29af78a960 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/mircounts commit d4d8106d66b65679a1a685ab94bfcf99cdb7b959
artbio
parents:
diff changeset
27
da29af78a960 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/mircounts commit d4d8106d66b65679a1a685ab94bfcf99cdb7b959
artbio
parents:
diff changeset
28 def get_pre_mir_counts(bamfile):
da29af78a960 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/mircounts commit d4d8106d66b65679a1a685ab94bfcf99cdb7b959
artbio
parents:
diff changeset
29 """
da29af78a960 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/mircounts commit d4d8106d66b65679a1a685ab94bfcf99cdb7b959
artbio
parents:
diff changeset
30 Takes a AlignmentFile object and returns a dictionary of counts for reads
da29af78a960 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/mircounts commit d4d8106d66b65679a1a685ab94bfcf99cdb7b959
artbio
parents:
diff changeset
31 aligning with pre_mirs (as keys)
da29af78a960 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/mircounts commit d4d8106d66b65679a1a685ab94bfcf99cdb7b959
artbio
parents:
diff changeset
32 """
da29af78a960 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/mircounts commit d4d8106d66b65679a1a685ab94bfcf99cdb7b959
artbio
parents:
diff changeset
33 count = dict()
da29af78a960 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/mircounts commit d4d8106d66b65679a1a685ab94bfcf99cdb7b959
artbio
parents:
diff changeset
34 for ref_name in bamfile.references:
da29af78a960 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/mircounts commit d4d8106d66b65679a1a685ab94bfcf99cdb7b959
artbio
parents:
diff changeset
35 count[ref_name] = bamfile.count(reference=ref_name)
da29af78a960 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/mircounts commit d4d8106d66b65679a1a685ab94bfcf99cdb7b959
artbio
parents:
diff changeset
36 return count
da29af78a960 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/mircounts commit d4d8106d66b65679a1a685ab94bfcf99cdb7b959
artbio
parents:
diff changeset
37
da29af78a960 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/mircounts commit d4d8106d66b65679a1a685ab94bfcf99cdb7b959
artbio
parents:
diff changeset
38
da29af78a960 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/mircounts commit d4d8106d66b65679a1a685ab94bfcf99cdb7b959
artbio
parents:
diff changeset
39 def get_pre_mir_coverage(bamfile, quality=10):
da29af78a960 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/mircounts commit d4d8106d66b65679a1a685ab94bfcf99cdb7b959
artbio
parents:
diff changeset
40 """
da29af78a960 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/mircounts commit d4d8106d66b65679a1a685ab94bfcf99cdb7b959
artbio
parents:
diff changeset
41 Takes a AlignmentFile object and returns a dictionary of lists
da29af78a960 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/mircounts commit d4d8106d66b65679a1a685ab94bfcf99cdb7b959
artbio
parents:
diff changeset
42 of coverage along the coordinates of pre_mirs (as keys)
da29af78a960 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/mircounts commit d4d8106d66b65679a1a685ab94bfcf99cdb7b959
artbio
parents:
diff changeset
43 """
da29af78a960 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/mircounts commit d4d8106d66b65679a1a685ab94bfcf99cdb7b959
artbio
parents:
diff changeset
44 coverage = dict()
da29af78a960 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/mircounts commit d4d8106d66b65679a1a685ab94bfcf99cdb7b959
artbio
parents:
diff changeset
45 for ref_name, ref_len in zip(bamfile.references, bamfile.lengths):
da29af78a960 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/mircounts commit d4d8106d66b65679a1a685ab94bfcf99cdb7b959
artbio
parents:
diff changeset
46 coverage[ref_name] = bamfile.count_coverage(reference=ref_name,
da29af78a960 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/mircounts commit d4d8106d66b65679a1a685ab94bfcf99cdb7b959
artbio
parents:
diff changeset
47 start=0, end=ref_len,
da29af78a960 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/mircounts commit d4d8106d66b65679a1a685ab94bfcf99cdb7b959
artbio
parents:
diff changeset
48 quality_threshold=quality)
da29af78a960 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/mircounts commit d4d8106d66b65679a1a685ab94bfcf99cdb7b959
artbio
parents:
diff changeset
49 """ Add the 4 coverage values """
da29af78a960 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/mircounts commit d4d8106d66b65679a1a685ab94bfcf99cdb7b959
artbio
parents:
diff changeset
50 coverage[ref_name] = [sum(x) for x in
da29af78a960 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/mircounts commit d4d8106d66b65679a1a685ab94bfcf99cdb7b959
artbio
parents:
diff changeset
51 zip(*coverage[ref_name])]
da29af78a960 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/mircounts commit d4d8106d66b65679a1a685ab94bfcf99cdb7b959
artbio
parents:
diff changeset
52 return coverage
da29af78a960 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/mircounts commit d4d8106d66b65679a1a685ab94bfcf99cdb7b959
artbio
parents:
diff changeset
53
da29af78a960 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/mircounts commit d4d8106d66b65679a1a685ab94bfcf99cdb7b959
artbio
parents:
diff changeset
54
da29af78a960 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/mircounts commit d4d8106d66b65679a1a685ab94bfcf99cdb7b959
artbio
parents:
diff changeset
55 def get_mir_counts(bamfile, gff_file):
da29af78a960 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/mircounts commit d4d8106d66b65679a1a685ab94bfcf99cdb7b959
artbio
parents:
diff changeset
56 """
da29af78a960 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/mircounts commit d4d8106d66b65679a1a685ab94bfcf99cdb7b959
artbio
parents:
diff changeset
57 Takes a AlignmentFile and a gff file and computes for
da29af78a960 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/mircounts commit d4d8106d66b65679a1a685ab94bfcf99cdb7b959
artbio
parents:
diff changeset
58 each 'miRNA' region of the gff the number of reads that hit it
da29af78a960 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/mircounts commit d4d8106d66b65679a1a685ab94bfcf99cdb7b959
artbio
parents:
diff changeset
59 returns a dict[mir_name] = count
da29af78a960 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/mircounts commit d4d8106d66b65679a1a685ab94bfcf99cdb7b959
artbio
parents:
diff changeset
60 """
da29af78a960 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/mircounts commit d4d8106d66b65679a1a685ab94bfcf99cdb7b959
artbio
parents:
diff changeset
61 counts = dict()
da29af78a960 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/mircounts commit d4d8106d66b65679a1a685ab94bfcf99cdb7b959
artbio
parents:
diff changeset
62 for line in open(gff_file, 'r'):
da29af78a960 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/mircounts commit d4d8106d66b65679a1a685ab94bfcf99cdb7b959
artbio
parents:
diff changeset
63 if line[0] != '#':
da29af78a960 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/mircounts commit d4d8106d66b65679a1a685ab94bfcf99cdb7b959
artbio
parents:
diff changeset
64 gff_fields = line[:-1].split("\t")
da29af78a960 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/mircounts commit d4d8106d66b65679a1a685ab94bfcf99cdb7b959
artbio
parents:
diff changeset
65 if gff_fields[2] == 'miRNA':
da29af78a960 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/mircounts commit d4d8106d66b65679a1a685ab94bfcf99cdb7b959
artbio
parents:
diff changeset
66 mir_name = gff_fields[0]
da29af78a960 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/mircounts commit d4d8106d66b65679a1a685ab94bfcf99cdb7b959
artbio
parents:
diff changeset
67 premir_name = gff_fields[8].split('=')[-1]
da29af78a960 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/mircounts commit d4d8106d66b65679a1a685ab94bfcf99cdb7b959
artbio
parents:
diff changeset
68 mir_start = int(gff_fields[3])
da29af78a960 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/mircounts commit d4d8106d66b65679a1a685ab94bfcf99cdb7b959
artbio
parents:
diff changeset
69 mir_end = int(gff_fields[4])
da29af78a960 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/mircounts commit d4d8106d66b65679a1a685ab94bfcf99cdb7b959
artbio
parents:
diff changeset
70 # GFF is 1-based, pysam is 0-based.
da29af78a960 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/mircounts commit d4d8106d66b65679a1a685ab94bfcf99cdb7b959
artbio
parents:
diff changeset
71 counts[mir_name] = bamfile.count(reference=premir_name,
da29af78a960 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/mircounts commit d4d8106d66b65679a1a685ab94bfcf99cdb7b959
artbio
parents:
diff changeset
72 start=mir_start-1,
da29af78a960 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/mircounts commit d4d8106d66b65679a1a685ab94bfcf99cdb7b959
artbio
parents:
diff changeset
73 end=mir_end-1)
da29af78a960 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/mircounts commit d4d8106d66b65679a1a685ab94bfcf99cdb7b959
artbio
parents:
diff changeset
74 return counts
da29af78a960 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/mircounts commit d4d8106d66b65679a1a685ab94bfcf99cdb7b959
artbio
parents:
diff changeset
75
da29af78a960 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/mircounts commit d4d8106d66b65679a1a685ab94bfcf99cdb7b959
artbio
parents:
diff changeset
76
da29af78a960 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/mircounts commit d4d8106d66b65679a1a685ab94bfcf99cdb7b959
artbio
parents:
diff changeset
77 def write_dataframe_coverage(countdict, outfile):
da29af78a960 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/mircounts commit d4d8106d66b65679a1a685ab94bfcf99cdb7b959
artbio
parents:
diff changeset
78 """
da29af78a960 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/mircounts commit d4d8106d66b65679a1a685ab94bfcf99cdb7b959
artbio
parents:
diff changeset
79 Takes a dict[pre_mir reference name] = [coverage list]
da29af78a960 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/mircounts commit d4d8106d66b65679a1a685ab94bfcf99cdb7b959
artbio
parents:
diff changeset
80 and writes a dataframe with columns:
da29af78a960 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/mircounts commit d4d8106d66b65679a1a685ab94bfcf99cdb7b959
artbio
parents:
diff changeset
81 <gene_type name>, offset, normoffset, counts and normcounts
da29af78a960 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/mircounts commit d4d8106d66b65679a1a685ab94bfcf99cdb7b959
artbio
parents:
diff changeset
82 in the outfile
da29af78a960 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/mircounts commit d4d8106d66b65679a1a685ab94bfcf99cdb7b959
artbio
parents:
diff changeset
83 """
da29af78a960 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/mircounts commit d4d8106d66b65679a1a685ab94bfcf99cdb7b959
artbio
parents:
diff changeset
84 F = open(outfile, 'w')
da29af78a960 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/mircounts commit d4d8106d66b65679a1a685ab94bfcf99cdb7b959
artbio
parents:
diff changeset
85 F.write('Mir_hairpin\tOffset\tNorm_offset\tCount\tNorm_count\n')
da29af78a960 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/mircounts commit d4d8106d66b65679a1a685ab94bfcf99cdb7b959
artbio
parents:
diff changeset
86 for ref in sorted(countdict):
da29af78a960 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/mircounts commit d4d8106d66b65679a1a685ab94bfcf99cdb7b959
artbio
parents:
diff changeset
87 """
da29af78a960 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/mircounts commit d4d8106d66b65679a1a685ab94bfcf99cdb7b959
artbio
parents:
diff changeset
88 For each reference name in mirs,
da29af78a960 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/mircounts commit d4d8106d66b65679a1a685ab94bfcf99cdb7b959
artbio
parents:
diff changeset
89 write the coverage of each of its positions
da29af78a960 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/mircounts commit d4d8106d66b65679a1a685ab94bfcf99cdb7b959
artbio
parents:
diff changeset
90 """
da29af78a960 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/mircounts commit d4d8106d66b65679a1a685ab94bfcf99cdb7b959
artbio
parents:
diff changeset
91 maximum = max(countdict[ref])
da29af78a960 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/mircounts commit d4d8106d66b65679a1a685ab94bfcf99cdb7b959
artbio
parents:
diff changeset
92 reference_length = len(countdict[ref])
da29af78a960 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/mircounts commit d4d8106d66b65679a1a685ab94bfcf99cdb7b959
artbio
parents:
diff changeset
93 for pos, c in enumerate(countdict[ref]):
da29af78a960 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/mircounts commit d4d8106d66b65679a1a685ab94bfcf99cdb7b959
artbio
parents:
diff changeset
94 """ Compute and write value for each reference position"""
da29af78a960 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/mircounts commit d4d8106d66b65679a1a685ab94bfcf99cdb7b959
artbio
parents:
diff changeset
95 F.write('%s\t%s\t%s\t%s\t%s\n' % (ref, str(pos + 1),
da29af78a960 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/mircounts commit d4d8106d66b65679a1a685ab94bfcf99cdb7b959
artbio
parents:
diff changeset
96 str(float(pos+1)/reference_length), str(float(c)),
da29af78a960 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/mircounts commit d4d8106d66b65679a1a685ab94bfcf99cdb7b959
artbio
parents:
diff changeset
97 str(float(c)/maximum) if maximum != 0 else '0'))
da29af78a960 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/mircounts commit d4d8106d66b65679a1a685ab94bfcf99cdb7b959
artbio
parents:
diff changeset
98 F.close()
da29af78a960 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/mircounts commit d4d8106d66b65679a1a685ab94bfcf99cdb7b959
artbio
parents:
diff changeset
99
da29af78a960 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/mircounts commit d4d8106d66b65679a1a685ab94bfcf99cdb7b959
artbio
parents:
diff changeset
100
da29af78a960 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/mircounts commit d4d8106d66b65679a1a685ab94bfcf99cdb7b959
artbio
parents:
diff changeset
101 def write_counts(countdict, outfile):
da29af78a960 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/mircounts commit d4d8106d66b65679a1a685ab94bfcf99cdb7b959
artbio
parents:
diff changeset
102 """
da29af78a960 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/mircounts commit d4d8106d66b65679a1a685ab94bfcf99cdb7b959
artbio
parents:
diff changeset
103 Takes a dict[<gene_type name>]=count and
da29af78a960 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/mircounts commit d4d8106d66b65679a1a685ab94bfcf99cdb7b959
artbio
parents:
diff changeset
104 writes a count table
da29af78a960 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/mircounts commit d4d8106d66b65679a1a685ab94bfcf99cdb7b959
artbio
parents:
diff changeset
105 """
da29af78a960 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/mircounts commit d4d8106d66b65679a1a685ab94bfcf99cdb7b959
artbio
parents:
diff changeset
106 F = open(outfile, 'w')
da29af78a960 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/mircounts commit d4d8106d66b65679a1a685ab94bfcf99cdb7b959
artbio
parents:
diff changeset
107 F.write('Gene\tCounts\n')
da29af78a960 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/mircounts commit d4d8106d66b65679a1a685ab94bfcf99cdb7b959
artbio
parents:
diff changeset
108 for gene in sorted(countdict):
da29af78a960 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/mircounts commit d4d8106d66b65679a1a685ab94bfcf99cdb7b959
artbio
parents:
diff changeset
109 F.write('%s\t%s\n' % (gene, str(countdict[gene])))
da29af78a960 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/mircounts commit d4d8106d66b65679a1a685ab94bfcf99cdb7b959
artbio
parents:
diff changeset
110 F.close()
da29af78a960 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/mircounts commit d4d8106d66b65679a1a685ab94bfcf99cdb7b959
artbio
parents:
diff changeset
111
da29af78a960 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/mircounts commit d4d8106d66b65679a1a685ab94bfcf99cdb7b959
artbio
parents:
diff changeset
112
da29af78a960 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/mircounts commit d4d8106d66b65679a1a685ab94bfcf99cdb7b959
artbio
parents:
diff changeset
113 def main():
da29af78a960 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/mircounts commit d4d8106d66b65679a1a685ab94bfcf99cdb7b959
artbio
parents:
diff changeset
114 args = Parser()
da29af78a960 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/mircounts commit d4d8106d66b65679a1a685ab94bfcf99cdb7b959
artbio
parents:
diff changeset
115 bamfile = pysam.AlignmentFile(args.alignment_file, 'rb', check_sq=False)
da29af78a960 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/mircounts commit d4d8106d66b65679a1a685ab94bfcf99cdb7b959
artbio
parents:
diff changeset
116 if args.pre_mirs:
da29af78a960 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/mircounts commit d4d8106d66b65679a1a685ab94bfcf99cdb7b959
artbio
parents:
diff changeset
117 pre_mirs = get_pre_mir_counts(bamfile)
da29af78a960 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/mircounts commit d4d8106d66b65679a1a685ab94bfcf99cdb7b959
artbio
parents:
diff changeset
118 write_counts(pre_mirs, args.pre_mirs)
da29af78a960 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/mircounts commit d4d8106d66b65679a1a685ab94bfcf99cdb7b959
artbio
parents:
diff changeset
119 if args.lattice:
da29af78a960 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/mircounts commit d4d8106d66b65679a1a685ab94bfcf99cdb7b959
artbio
parents:
diff changeset
120 pre_mirs_coverage = get_pre_mir_coverage(bamfile,
da29af78a960 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/mircounts commit d4d8106d66b65679a1a685ab94bfcf99cdb7b959
artbio
parents:
diff changeset
121 args.quality_threshold)
da29af78a960 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/mircounts commit d4d8106d66b65679a1a685ab94bfcf99cdb7b959
artbio
parents:
diff changeset
122 write_dataframe_coverage(pre_mirs_coverage, args.lattice)
da29af78a960 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/mircounts commit d4d8106d66b65679a1a685ab94bfcf99cdb7b959
artbio
parents:
diff changeset
123 if args.mirs:
da29af78a960 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/mircounts commit d4d8106d66b65679a1a685ab94bfcf99cdb7b959
artbio
parents:
diff changeset
124 mirs = get_mir_counts(bamfile, args.gff_file)
da29af78a960 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/mircounts commit d4d8106d66b65679a1a685ab94bfcf99cdb7b959
artbio
parents:
diff changeset
125 write_counts(mirs, args.mirs)
da29af78a960 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/mircounts commit d4d8106d66b65679a1a685ab94bfcf99cdb7b959
artbio
parents:
diff changeset
126
da29af78a960 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/mircounts commit d4d8106d66b65679a1a685ab94bfcf99cdb7b959
artbio
parents:
diff changeset
127
da29af78a960 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/mircounts commit d4d8106d66b65679a1a685ab94bfcf99cdb7b959
artbio
parents:
diff changeset
128 if __name__ == '__main__':
da29af78a960 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/mircounts commit d4d8106d66b65679a1a685ab94bfcf99cdb7b959
artbio
parents:
diff changeset
129 main()