Mercurial > repos > artbio > mircounts

annotate mircounts.py @ 4:b7ac138bb781 draft default tip

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/mircounts commit 51dc6c56c7d95fc229ffee958354211cd454fd36"
author artbio
date Sun, 09 May 2021 17:06:45 +0000
parents ee99c6374a3b
children
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
rev   line source
0
10f0e4c00b13 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/mircounts commit 3a6181bd181729f642b75c4e689f063fc2821cf1
artbio
parents:
diff changeset
1 #!/usr/bin/env python
10f0e4c00b13 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/mircounts commit 3a6181bd181729f642b75c4e689f063fc2821cf1
artbio
parents:
diff changeset
2 import argparse
10f0e4c00b13 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/mircounts commit 3a6181bd181729f642b75c4e689f063fc2821cf1
artbio
parents:
diff changeset
3
10f0e4c00b13 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/mircounts commit 3a6181bd181729f642b75c4e689f063fc2821cf1
artbio
parents:
diff changeset
4 import pysam
10f0e4c00b13 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/mircounts commit 3a6181bd181729f642b75c4e689f063fc2821cf1
artbio
parents:
diff changeset
5
10f0e4c00b13 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/mircounts commit 3a6181bd181729f642b75c4e689f063fc2821cf1
artbio
parents:
diff changeset
6
10f0e4c00b13 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/mircounts commit 3a6181bd181729f642b75c4e689f063fc2821cf1
artbio
parents:
diff changeset
7 def Parser():
10f0e4c00b13 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/mircounts commit 3a6181bd181729f642b75c4e689f063fc2821cf1
artbio
parents:
diff changeset
8 parser = argparse.ArgumentParser(description='miRNAs counts and coverages')
10f0e4c00b13 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/mircounts commit 3a6181bd181729f642b75c4e689f063fc2821cf1
artbio
parents:
diff changeset
9 parser.add_argument('-a', '--alignment', metavar='FILE', type=str,
10f0e4c00b13 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/mircounts commit 3a6181bd181729f642b75c4e689f063fc2821cf1
artbio
parents:
diff changeset
10 dest='alignment_file', help='Alignment bam file')
10f0e4c00b13 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/mircounts commit 3a6181bd181729f642b75c4e689f063fc2821cf1
artbio
parents:
diff changeset
11 parser.add_argument('--gff', metavar='FILE', type=str, dest='gff_file',
10f0e4c00b13 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/mircounts commit 3a6181bd181729f642b75c4e689f063fc2821cf1
artbio
parents:
diff changeset
12 help='GFF3 describing both pre-miRNAs\
10f0e4c00b13 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/mircounts commit 3a6181bd181729f642b75c4e689f063fc2821cf1
artbio
parents:
diff changeset
13 and mature miRNAs')
10f0e4c00b13 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/mircounts commit 3a6181bd181729f642b75c4e689f063fc2821cf1
artbio
parents:
diff changeset
14 parser.add_argument('-q', '--quality_threshold', type=int,
10f0e4c00b13 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/mircounts commit 3a6181bd181729f642b75c4e689f063fc2821cf1
artbio
parents:
diff changeset
15 dest='quality_threshold',
10f0e4c00b13 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/mircounts commit 3a6181bd181729f642b75c4e689f063fc2821cf1
artbio
parents:
diff changeset
16 help='Quality threshold for coverage (default=10)',
10f0e4c00b13 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/mircounts commit 3a6181bd181729f642b75c4e689f063fc2821cf1
artbio
parents:
diff changeset
17 default=10)
10f0e4c00b13 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/mircounts commit 3a6181bd181729f642b75c4e689f063fc2821cf1
artbio
parents:
diff changeset
18 parser.add_argument('-p', '--pre_mirs', type=str, dest='pre_mirs',
10f0e4c00b13 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/mircounts commit 3a6181bd181729f642b75c4e689f063fc2821cf1
artbio
parents:
diff changeset
19 help='pre-miRNAs count file path', metavar='FILE')
10f0e4c00b13 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/mircounts commit 3a6181bd181729f642b75c4e689f063fc2821cf1
artbio
parents:
diff changeset
20 parser.add_argument('-m', '--mirs', type=str, dest='mirs',
10f0e4c00b13 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/mircounts commit 3a6181bd181729f642b75c4e689f063fc2821cf1
artbio
parents:
diff changeset
21 help='mature miRNA count file path', metavar='FILE')
10f0e4c00b13 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/mircounts commit 3a6181bd181729f642b75c4e689f063fc2821cf1
artbio
parents:
diff changeset
22 parser.add_argument('--lattice', metavar='FILE', type=str, dest='lattice',
10f0e4c00b13 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/mircounts commit 3a6181bd181729f642b75c4e689f063fc2821cf1
artbio
parents:
diff changeset
23 help='Output file for the lattice dataframe.')
10f0e4c00b13 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/mircounts commit 3a6181bd181729f642b75c4e689f063fc2821cf1
artbio
parents:
diff changeset
24 args = parser.parse_args()
10f0e4c00b13 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/mircounts commit 3a6181bd181729f642b75c4e689f063fc2821cf1
artbio
parents:
diff changeset
25 return args
10f0e4c00b13 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/mircounts commit 3a6181bd181729f642b75c4e689f063fc2821cf1
artbio
parents:
diff changeset
26
10f0e4c00b13 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/mircounts commit 3a6181bd181729f642b75c4e689f063fc2821cf1
artbio
parents:
diff changeset
27
10f0e4c00b13 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/mircounts commit 3a6181bd181729f642b75c4e689f063fc2821cf1
artbio
parents:
diff changeset
28 def get_pre_mir_counts(bamfile):
10f0e4c00b13 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/mircounts commit 3a6181bd181729f642b75c4e689f063fc2821cf1
artbio
parents:
diff changeset
29 """
10f0e4c00b13 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/mircounts commit 3a6181bd181729f642b75c4e689f063fc2821cf1
artbio
parents:
diff changeset
30 Takes a AlignmentFile object and returns a dictionary of counts for reads
10f0e4c00b13 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/mircounts commit 3a6181bd181729f642b75c4e689f063fc2821cf1
artbio
parents:
diff changeset
31 aligning with pre_mirs (as keys)
10f0e4c00b13 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/mircounts commit 3a6181bd181729f642b75c4e689f063fc2821cf1
artbio
parents:
diff changeset
32 """
10f0e4c00b13 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/mircounts commit 3a6181bd181729f642b75c4e689f063fc2821cf1
artbio
parents:
diff changeset
33 count = dict()
10f0e4c00b13 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/mircounts commit 3a6181bd181729f642b75c4e689f063fc2821cf1
artbio
parents:
diff changeset
34 for ref_name in bamfile.references:
10f0e4c00b13 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/mircounts commit 3a6181bd181729f642b75c4e689f063fc2821cf1
artbio
parents:
diff changeset
35 count[ref_name] = bamfile.count(reference=ref_name)
10f0e4c00b13 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/mircounts commit 3a6181bd181729f642b75c4e689f063fc2821cf1
artbio
parents:
diff changeset
36 return count
10f0e4c00b13 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/mircounts commit 3a6181bd181729f642b75c4e689f063fc2821cf1
artbio
parents:
diff changeset
37
10f0e4c00b13 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/mircounts commit 3a6181bd181729f642b75c4e689f063fc2821cf1
artbio
parents:
diff changeset
38
10f0e4c00b13 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/mircounts commit 3a6181bd181729f642b75c4e689f063fc2821cf1
artbio
parents:
diff changeset
39 def get_pre_mir_coverage(bamfile, quality=10):
10f0e4c00b13 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/mircounts commit 3a6181bd181729f642b75c4e689f063fc2821cf1
artbio
parents:
diff changeset
40 """
10f0e4c00b13 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/mircounts commit 3a6181bd181729f642b75c4e689f063fc2821cf1
artbio
parents:
diff changeset
41 Takes a AlignmentFile object and returns a dictionary of lists
10f0e4c00b13 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/mircounts commit 3a6181bd181729f642b75c4e689f063fc2821cf1
artbio
parents:
diff changeset
42 of coverage along the coordinates of pre_mirs (as keys)
10f0e4c00b13 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/mircounts commit 3a6181bd181729f642b75c4e689f063fc2821cf1
artbio
parents:
diff changeset
43 """
10f0e4c00b13 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/mircounts commit 3a6181bd181729f642b75c4e689f063fc2821cf1
artbio
parents:
diff changeset
44 coverage = dict()
10f0e4c00b13 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/mircounts commit 3a6181bd181729f642b75c4e689f063fc2821cf1
artbio
parents:
diff changeset
45 for ref_name, ref_len in zip(bamfile.references, bamfile.lengths):
3
ee99c6374a3b "planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/mircounts commit af48e9f6df2717ffd3731a974be1ec36e4eff779"
artbio
parents: 0
diff changeset
46 coverage[ref_name] = bamfile.count_coverage(contig=ref_name,
ee99c6374a3b "planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/mircounts commit af48e9f6df2717ffd3731a974be1ec36e4eff779"
artbio
parents: 0
diff changeset
47 start=0, stop=ref_len,
0
10f0e4c00b13 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/mircounts commit 3a6181bd181729f642b75c4e689f063fc2821cf1
artbio
parents:
diff changeset
48 quality_threshold=quality)
10f0e4c00b13 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/mircounts commit 3a6181bd181729f642b75c4e689f063fc2821cf1
artbio
parents:
diff changeset
49 """ Add the 4 coverage values """
10f0e4c00b13 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/mircounts commit 3a6181bd181729f642b75c4e689f063fc2821cf1
artbio
parents:
diff changeset
50 coverage[ref_name] = [sum(x) for x in
10f0e4c00b13 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/mircounts commit 3a6181bd181729f642b75c4e689f063fc2821cf1
artbio
parents:
diff changeset
51 zip(*coverage[ref_name])]
10f0e4c00b13 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/mircounts commit 3a6181bd181729f642b75c4e689f063fc2821cf1
artbio
parents:
diff changeset
52 return coverage
10f0e4c00b13 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/mircounts commit 3a6181bd181729f642b75c4e689f063fc2821cf1
artbio
parents:
diff changeset
53
10f0e4c00b13 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/mircounts commit 3a6181bd181729f642b75c4e689f063fc2821cf1
artbio
parents:
diff changeset
54
10f0e4c00b13 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/mircounts commit 3a6181bd181729f642b75c4e689f063fc2821cf1
artbio
parents:
diff changeset
55 def get_mir_counts(bamfile, gff_file):
10f0e4c00b13 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/mircounts commit 3a6181bd181729f642b75c4e689f063fc2821cf1
artbio
parents:
diff changeset
56 """
10f0e4c00b13 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/mircounts commit 3a6181bd181729f642b75c4e689f063fc2821cf1
artbio
parents:
diff changeset
57 Takes a AlignmentFile and a gff file and computes for
10f0e4c00b13 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/mircounts commit 3a6181bd181729f642b75c4e689f063fc2821cf1
artbio
parents:
diff changeset
58 each 'miRNA' region of the gff the number of reads that hit it
10f0e4c00b13 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/mircounts commit 3a6181bd181729f642b75c4e689f063fc2821cf1
artbio
parents:
diff changeset
59 returns a dict[mir_name] = count
10f0e4c00b13 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/mircounts commit 3a6181bd181729f642b75c4e689f063fc2821cf1
artbio
parents:
diff changeset
60 """
10f0e4c00b13 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/mircounts commit 3a6181bd181729f642b75c4e689f063fc2821cf1
artbio
parents:
diff changeset
61 counts = dict()
10f0e4c00b13 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/mircounts commit 3a6181bd181729f642b75c4e689f063fc2821cf1
artbio
parents:
diff changeset
62 for line in open(gff_file, 'r'):
10f0e4c00b13 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/mircounts commit 3a6181bd181729f642b75c4e689f063fc2821cf1
artbio
parents:
diff changeset
63 if line[0] != '#':
10f0e4c00b13 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/mircounts commit 3a6181bd181729f642b75c4e689f063fc2821cf1
artbio
parents:
diff changeset
64 gff_fields = line[:-1].split("\t")
10f0e4c00b13 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/mircounts commit 3a6181bd181729f642b75c4e689f063fc2821cf1
artbio
parents:
diff changeset
65 if gff_fields[2] == 'miRNA':
10f0e4c00b13 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/mircounts commit 3a6181bd181729f642b75c4e689f063fc2821cf1
artbio
parents:
diff changeset
66 mir_name = gff_fields[0]
3
ee99c6374a3b "planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/mircounts commit af48e9f6df2717ffd3731a974be1ec36e4eff779"
artbio
parents: 0
diff changeset
67 premir_name = gff_fields[8].split('Parent_mir_Name=')[-1]
0
10f0e4c00b13 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/mircounts commit 3a6181bd181729f642b75c4e689f063fc2821cf1
artbio
parents:
diff changeset
68 mir_start = int(gff_fields[3])
10f0e4c00b13 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/mircounts commit 3a6181bd181729f642b75c4e689f063fc2821cf1
artbio
parents:
diff changeset
69 mir_end = int(gff_fields[4])
10f0e4c00b13 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/mircounts commit 3a6181bd181729f642b75c4e689f063fc2821cf1
artbio
parents:
diff changeset
70 # GFF is 1-based, pysam is 0-based.
3
ee99c6374a3b "planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/mircounts commit af48e9f6df2717ffd3731a974be1ec36e4eff779"
artbio
parents: 0
diff changeset
71 counts[mir_name] = bamfile.count(contig=premir_name,
0
10f0e4c00b13 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/mircounts commit 3a6181bd181729f642b75c4e689f063fc2821cf1
artbio
parents:
diff changeset
72 start=mir_start-1,
3
ee99c6374a3b "planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/mircounts commit af48e9f6df2717ffd3731a974be1ec36e4eff779"
artbio
parents: 0
diff changeset
73 stop=mir_end-1)
0
10f0e4c00b13 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/mircounts commit 3a6181bd181729f642b75c4e689f063fc2821cf1
artbio
parents:
diff changeset
74 return counts
10f0e4c00b13 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/mircounts commit 3a6181bd181729f642b75c4e689f063fc2821cf1
artbio
parents:
diff changeset
75
10f0e4c00b13 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/mircounts commit 3a6181bd181729f642b75c4e689f063fc2821cf1
artbio
parents:
diff changeset
76
10f0e4c00b13 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/mircounts commit 3a6181bd181729f642b75c4e689f063fc2821cf1
artbio
parents:
diff changeset
77 def write_dataframe_coverage(countdict, outfile):
10f0e4c00b13 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/mircounts commit 3a6181bd181729f642b75c4e689f063fc2821cf1
artbio
parents:
diff changeset
78 """
10f0e4c00b13 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/mircounts commit 3a6181bd181729f642b75c4e689f063fc2821cf1
artbio
parents:
diff changeset
79 Takes a dict[pre_mir reference name] = [coverage list]
10f0e4c00b13 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/mircounts commit 3a6181bd181729f642b75c4e689f063fc2821cf1
artbio
parents:
diff changeset
80 and writes a dataframe with columns:
10f0e4c00b13 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/mircounts commit 3a6181bd181729f642b75c4e689f063fc2821cf1
artbio
parents:
diff changeset
81 <gene_type name>, offset, normoffset, counts and normcounts
10f0e4c00b13 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/mircounts commit 3a6181bd181729f642b75c4e689f063fc2821cf1
artbio
parents:
diff changeset
82 in the outfile
10f0e4c00b13 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/mircounts commit 3a6181bd181729f642b75c4e689f063fc2821cf1
artbio
parents:
diff changeset
83 """
10f0e4c00b13 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/mircounts commit 3a6181bd181729f642b75c4e689f063fc2821cf1
artbio
parents:
diff changeset
84 F = open(outfile, 'w')
10f0e4c00b13 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/mircounts commit 3a6181bd181729f642b75c4e689f063fc2821cf1
artbio
parents:
diff changeset
85 F.write('Mir_hairpin\tOffset\tNorm_offset\tCount\tNorm_count\n')
10f0e4c00b13 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/mircounts commit 3a6181bd181729f642b75c4e689f063fc2821cf1
artbio
parents:
diff changeset
86 for ref in sorted(countdict):
10f0e4c00b13 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/mircounts commit 3a6181bd181729f642b75c4e689f063fc2821cf1
artbio
parents:
diff changeset
87 """
10f0e4c00b13 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/mircounts commit 3a6181bd181729f642b75c4e689f063fc2821cf1
artbio
parents:
diff changeset
88 For each reference name in mirs,
10f0e4c00b13 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/mircounts commit 3a6181bd181729f642b75c4e689f063fc2821cf1
artbio
parents:
diff changeset
89 write the coverage of each of its positions
10f0e4c00b13 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/mircounts commit 3a6181bd181729f642b75c4e689f063fc2821cf1
artbio
parents:
diff changeset
90 """
10f0e4c00b13 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/mircounts commit 3a6181bd181729f642b75c4e689f063fc2821cf1
artbio
parents:
diff changeset
91 maximum = max(countdict[ref])
10f0e4c00b13 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/mircounts commit 3a6181bd181729f642b75c4e689f063fc2821cf1
artbio
parents:
diff changeset
92 reference_length = len(countdict[ref])
10f0e4c00b13 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/mircounts commit 3a6181bd181729f642b75c4e689f063fc2821cf1
artbio
parents:
diff changeset
93 for pos, c in enumerate(countdict[ref]):
10f0e4c00b13 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/mircounts commit 3a6181bd181729f642b75c4e689f063fc2821cf1
artbio
parents:
diff changeset
94 """ Compute and write value for each reference position"""
10f0e4c00b13 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/mircounts commit 3a6181bd181729f642b75c4e689f063fc2821cf1
artbio
parents:
diff changeset
95 F.write('%s\t%s\t%s\t%s\t%s\n' % (ref, str(pos + 1),
10f0e4c00b13 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/mircounts commit 3a6181bd181729f642b75c4e689f063fc2821cf1
artbio
parents:
diff changeset
96 str(float(pos+1)/reference_length), str(float(c)),
10f0e4c00b13 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/mircounts commit 3a6181bd181729f642b75c4e689f063fc2821cf1
artbio
parents:
diff changeset
97 str(float(c)/maximum) if maximum != 0 else '0'))
10f0e4c00b13 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/mircounts commit 3a6181bd181729f642b75c4e689f063fc2821cf1
artbio
parents:
diff changeset
98 F.close()
10f0e4c00b13 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/mircounts commit 3a6181bd181729f642b75c4e689f063fc2821cf1
artbio
parents:
diff changeset
99
10f0e4c00b13 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/mircounts commit 3a6181bd181729f642b75c4e689f063fc2821cf1
artbio
parents:
diff changeset
100
10f0e4c00b13 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/mircounts commit 3a6181bd181729f642b75c4e689f063fc2821cf1
artbio
parents:
diff changeset
101 def write_counts(countdict, outfile):
10f0e4c00b13 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/mircounts commit 3a6181bd181729f642b75c4e689f063fc2821cf1
artbio
parents:
diff changeset
102 """
10f0e4c00b13 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/mircounts commit 3a6181bd181729f642b75c4e689f063fc2821cf1
artbio
parents:
diff changeset
103 Takes a dict[<gene_type name>]=count and
10f0e4c00b13 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/mircounts commit 3a6181bd181729f642b75c4e689f063fc2821cf1
artbio
parents:
diff changeset
104 writes a count table
10f0e4c00b13 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/mircounts commit 3a6181bd181729f642b75c4e689f063fc2821cf1
artbio
parents:
diff changeset
105 """
10f0e4c00b13 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/mircounts commit 3a6181bd181729f642b75c4e689f063fc2821cf1
artbio
parents:
diff changeset
106 F = open(outfile, 'w')
10f0e4c00b13 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/mircounts commit 3a6181bd181729f642b75c4e689f063fc2821cf1
artbio
parents:
diff changeset
107 for gene in sorted(countdict):
10f0e4c00b13 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/mircounts commit 3a6181bd181729f642b75c4e689f063fc2821cf1
artbio
parents:
diff changeset
108 F.write('%s\t%s\n' % (gene, str(countdict[gene])))
10f0e4c00b13 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/mircounts commit 3a6181bd181729f642b75c4e689f063fc2821cf1
artbio
parents:
diff changeset
109 F.close()
10f0e4c00b13 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/mircounts commit 3a6181bd181729f642b75c4e689f063fc2821cf1
artbio
parents:
diff changeset
110
10f0e4c00b13 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/mircounts commit 3a6181bd181729f642b75c4e689f063fc2821cf1
artbio
parents:
diff changeset
111
10f0e4c00b13 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/mircounts commit 3a6181bd181729f642b75c4e689f063fc2821cf1
artbio
parents:
diff changeset
112 def main():
10f0e4c00b13 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/mircounts commit 3a6181bd181729f642b75c4e689f063fc2821cf1
artbio
parents:
diff changeset
113 args = Parser()
10f0e4c00b13 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/mircounts commit 3a6181bd181729f642b75c4e689f063fc2821cf1
artbio
parents:
diff changeset
114 bamfile = pysam.AlignmentFile(args.alignment_file, 'rb', check_sq=False)
10f0e4c00b13 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/mircounts commit 3a6181bd181729f642b75c4e689f063fc2821cf1
artbio
parents:
diff changeset
115 if args.pre_mirs:
10f0e4c00b13 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/mircounts commit 3a6181bd181729f642b75c4e689f063fc2821cf1
artbio
parents:
diff changeset
116 pre_mirs = get_pre_mir_counts(bamfile)
10f0e4c00b13 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/mircounts commit 3a6181bd181729f642b75c4e689f063fc2821cf1
artbio
parents:
diff changeset
117 write_counts(pre_mirs, args.pre_mirs)
10f0e4c00b13 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/mircounts commit 3a6181bd181729f642b75c4e689f063fc2821cf1
artbio
parents:
diff changeset
118 if args.lattice:
10f0e4c00b13 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/mircounts commit 3a6181bd181729f642b75c4e689f063fc2821cf1
artbio
parents:
diff changeset
119 pre_mirs_coverage = get_pre_mir_coverage(bamfile,
10f0e4c00b13 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/mircounts commit 3a6181bd181729f642b75c4e689f063fc2821cf1
artbio
parents:
diff changeset
120 args.quality_threshold)
10f0e4c00b13 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/mircounts commit 3a6181bd181729f642b75c4e689f063fc2821cf1
artbio
parents:
diff changeset
121 write_dataframe_coverage(pre_mirs_coverage, args.lattice)
10f0e4c00b13 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/mircounts commit 3a6181bd181729f642b75c4e689f063fc2821cf1
artbio
parents:
diff changeset
122 if args.mirs:
10f0e4c00b13 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/mircounts commit 3a6181bd181729f642b75c4e689f063fc2821cf1
artbio
parents:
diff changeset
123 mirs = get_mir_counts(bamfile, args.gff_file)
10f0e4c00b13 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/mircounts commit 3a6181bd181729f642b75c4e689f063fc2821cf1
artbio
parents:
diff changeset
124 write_counts(mirs, args.mirs)
10f0e4c00b13 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/mircounts commit 3a6181bd181729f642b75c4e689f063fc2821cf1
artbio
parents:
diff changeset
125
10f0e4c00b13 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/mircounts commit 3a6181bd181729f642b75c4e689f063fc2821cf1
artbio
parents:
diff changeset
126
10f0e4c00b13 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/mircounts commit 3a6181bd181729f642b75c4e689f063fc2821cf1
artbio
parents:
diff changeset
127 if __name__ == '__main__':
10f0e4c00b13 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/mircounts commit 3a6181bd181729f642b75c4e689f063fc2821cf1
artbio
parents:
diff changeset
128 main()