annotate sequence_format_converter.py @ 4:1c14d2869d4d draft default tip

"planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/sequence_format_converter commit 9c5f0b8e89dfe4347c610f42923f0acad2ecc81b"
author artbio
date Wed, 17 Mar 2021 22:09:13 +0000
parents 772bd67ef26a
children
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
rev   line source
0
a8aacccd79a3 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/sequence_format_converter commit d6ef80f9db43eae4f58b33f58b5ef6f8209907db
artbio
parents:
diff changeset
1 #!/usr/bin/env python
a8aacccd79a3 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/sequence_format_converter commit d6ef80f9db43eae4f58b33f58b5ef6f8209907db
artbio
parents:
diff changeset
2 #
a8aacccd79a3 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/sequence_format_converter commit d6ef80f9db43eae4f58b33f58b5ef6f8209907db
artbio
parents:
diff changeset
3 import argparse
a8aacccd79a3 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/sequence_format_converter commit d6ef80f9db43eae4f58b33f58b5ef6f8209907db
artbio
parents:
diff changeset
4 import logging
a8aacccd79a3 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/sequence_format_converter commit d6ef80f9db43eae4f58b33f58b5ef6f8209907db
artbio
parents:
diff changeset
5 import sys
a8aacccd79a3 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/sequence_format_converter commit d6ef80f9db43eae4f58b33f58b5ef6f8209907db
artbio
parents:
diff changeset
6 from collections import defaultdict
a8aacccd79a3 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/sequence_format_converter commit d6ef80f9db43eae4f58b33f58b5ef6f8209907db
artbio
parents:
diff changeset
7
a8aacccd79a3 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/sequence_format_converter commit d6ef80f9db43eae4f58b33f58b5ef6f8209907db
artbio
parents:
diff changeset
8
a8aacccd79a3 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/sequence_format_converter commit d6ef80f9db43eae4f58b33f58b5ef6f8209907db
artbio
parents:
diff changeset
9 def Parser():
a8aacccd79a3 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/sequence_format_converter commit d6ef80f9db43eae4f58b33f58b5ef6f8209907db
artbio
parents:
diff changeset
10 the_parser = argparse.ArgumentParser()
a8aacccd79a3 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/sequence_format_converter commit d6ef80f9db43eae4f58b33f58b5ef6f8209907db
artbio
parents:
diff changeset
11 the_parser.add_argument(
a8aacccd79a3 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/sequence_format_converter commit d6ef80f9db43eae4f58b33f58b5ef6f8209907db
artbio
parents:
diff changeset
12 '--input', action="store", type=str,
a8aacccd79a3 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/sequence_format_converter commit d6ef80f9db43eae4f58b33f58b5ef6f8209907db
artbio
parents:
diff changeset
13 help="input file, accepted format: fastq, fasta, fasta_weigthed, \
a8aacccd79a3 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/sequence_format_converter commit d6ef80f9db43eae4f58b33f58b5ef6f8209907db
artbio
parents:
diff changeset
14 tabular")
a8aacccd79a3 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/sequence_format_converter commit d6ef80f9db43eae4f58b33f58b5ef6f8209907db
artbio
parents:
diff changeset
15 the_parser.add_argument(
a8aacccd79a3 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/sequence_format_converter commit d6ef80f9db43eae4f58b33f58b5ef6f8209907db
artbio
parents:
diff changeset
16 '--output', action="store", type=str, help="output converted file")
a8aacccd79a3 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/sequence_format_converter commit d6ef80f9db43eae4f58b33f58b5ef6f8209907db
artbio
parents:
diff changeset
17 the_parser.add_argument(
a8aacccd79a3 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/sequence_format_converter commit d6ef80f9db43eae4f58b33f58b5ef6f8209907db
artbio
parents:
diff changeset
18 '--format', action="store", type=str,
a8aacccd79a3 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/sequence_format_converter commit d6ef80f9db43eae4f58b33f58b5ef6f8209907db
artbio
parents:
diff changeset
19 help="select output format (fasta, fasta_weigthed, tabular")
a8aacccd79a3 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/sequence_format_converter commit d6ef80f9db43eae4f58b33f58b5ef6f8209907db
artbio
parents:
diff changeset
20 args = the_parser.parse_args()
a8aacccd79a3 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/sequence_format_converter commit d6ef80f9db43eae4f58b33f58b5ef6f8209907db
artbio
parents:
diff changeset
21 return args
a8aacccd79a3 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/sequence_format_converter commit d6ef80f9db43eae4f58b33f58b5ef6f8209907db
artbio
parents:
diff changeset
22
a8aacccd79a3 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/sequence_format_converter commit d6ef80f9db43eae4f58b33f58b5ef6f8209907db
artbio
parents:
diff changeset
23
a8aacccd79a3 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/sequence_format_converter commit d6ef80f9db43eae4f58b33f58b5ef6f8209907db
artbio
parents:
diff changeset
24 class Sequencing:
a8aacccd79a3 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/sequence_format_converter commit d6ef80f9db43eae4f58b33f58b5ef6f8209907db
artbio
parents:
diff changeset
25
a8aacccd79a3 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/sequence_format_converter commit d6ef80f9db43eae4f58b33f58b5ef6f8209907db
artbio
parents:
diff changeset
26 def __init__(self, input, output, format):
a8aacccd79a3 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/sequence_format_converter commit d6ef80f9db43eae4f58b33f58b5ef6f8209907db
artbio
parents:
diff changeset
27 self.input = input
a8aacccd79a3 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/sequence_format_converter commit d6ef80f9db43eae4f58b33f58b5ef6f8209907db
artbio
parents:
diff changeset
28 self.output = open(output, 'w')
a8aacccd79a3 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/sequence_format_converter commit d6ef80f9db43eae4f58b33f58b5ef6f8209907db
artbio
parents:
diff changeset
29 self.outputformat = format
a8aacccd79a3 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/sequence_format_converter commit d6ef80f9db43eae4f58b33f58b5ef6f8209907db
artbio
parents:
diff changeset
30 self.inputformat = self.detectformat(self.input)
a8aacccd79a3 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/sequence_format_converter commit d6ef80f9db43eae4f58b33f58b5ef6f8209907db
artbio
parents:
diff changeset
31 self.seqdic = defaultdict(int)
a8aacccd79a3 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/sequence_format_converter commit d6ef80f9db43eae4f58b33f58b5ef6f8209907db
artbio
parents:
diff changeset
32 self.read(self.input, self.inputformat)
a8aacccd79a3 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/sequence_format_converter commit d6ef80f9db43eae4f58b33f58b5ef6f8209907db
artbio
parents:
diff changeset
33 self.write(self.output, self.outputformat)
a8aacccd79a3 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/sequence_format_converter commit d6ef80f9db43eae4f58b33f58b5ef6f8209907db
artbio
parents:
diff changeset
34
a8aacccd79a3 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/sequence_format_converter commit d6ef80f9db43eae4f58b33f58b5ef6f8209907db
artbio
parents:
diff changeset
35 def detectformat(self, input):
a8aacccd79a3 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/sequence_format_converter commit d6ef80f9db43eae4f58b33f58b5ef6f8209907db
artbio
parents:
diff changeset
36 input = open(input, 'r')
a8aacccd79a3 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/sequence_format_converter commit d6ef80f9db43eae4f58b33f58b5ef6f8209907db
artbio
parents:
diff changeset
37 block = []
a8aacccd79a3 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/sequence_format_converter commit d6ef80f9db43eae4f58b33f58b5ef6f8209907db
artbio
parents:
diff changeset
38 reference = ['A', 'T', 'G', 'C', 'N']
a8aacccd79a3 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/sequence_format_converter commit d6ef80f9db43eae4f58b33f58b5ef6f8209907db
artbio
parents:
diff changeset
39 format = ''
a8aacccd79a3 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/sequence_format_converter commit d6ef80f9db43eae4f58b33f58b5ef6f8209907db
artbio
parents:
diff changeset
40 try:
3
772bd67ef26a "planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/sequence_format_converter commit 07c73df696d6d80e03f03232603d713882131625"
artbio
parents: 1
diff changeset
41 for n in range(4):
0
a8aacccd79a3 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/sequence_format_converter commit d6ef80f9db43eae4f58b33f58b5ef6f8209907db
artbio
parents:
diff changeset
42 block.append(input.readline()[:-1])
1
9ce7ccd468aa planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/sequence_format_converter commit dfecfb40f245a3cdb09dd1cfe37be4cb164ad2eb
artbio
parents: 0
diff changeset
43 except IndexError:
0
a8aacccd79a3 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/sequence_format_converter commit d6ef80f9db43eae4f58b33f58b5ef6f8209907db
artbio
parents:
diff changeset
44 logging.info("File hasn't at leat four lines !")
a8aacccd79a3 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/sequence_format_converter commit d6ef80f9db43eae4f58b33f58b5ef6f8209907db
artbio
parents:
diff changeset
45 sys.exit("File hasn't at leat four lines !")
a8aacccd79a3 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/sequence_format_converter commit d6ef80f9db43eae4f58b33f58b5ef6f8209907db
artbio
parents:
diff changeset
46 input.close()
a8aacccd79a3 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/sequence_format_converter commit d6ef80f9db43eae4f58b33f58b5ef6f8209907db
artbio
parents:
diff changeset
47 line1, line2, line3, line4 = block[0], block[1], block[2], block[3]
a8aacccd79a3 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/sequence_format_converter commit d6ef80f9db43eae4f58b33f58b5ef6f8209907db
artbio
parents:
diff changeset
48 if line1[0] == '>' and line3[0] == '>':
a8aacccd79a3 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/sequence_format_converter commit d6ef80f9db43eae4f58b33f58b5ef6f8209907db
artbio
parents:
diff changeset
49 logging.info("'>' detected in lines 1 and 3")
a8aacccd79a3 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/sequence_format_converter commit d6ef80f9db43eae4f58b33f58b5ef6f8209907db
artbio
parents:
diff changeset
50 sequence = ''.join([line2, line4]).upper()
a8aacccd79a3 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/sequence_format_converter commit d6ef80f9db43eae4f58b33f58b5ef6f8209907db
artbio
parents:
diff changeset
51 nucleotides = set([base for base in sequence])
a8aacccd79a3 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/sequence_format_converter commit d6ef80f9db43eae4f58b33f58b5ef6f8209907db
artbio
parents:
diff changeset
52 for nucleotide in nucleotides:
a8aacccd79a3 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/sequence_format_converter commit d6ef80f9db43eae4f58b33f58b5ef6f8209907db
artbio
parents:
diff changeset
53 if nucleotide not in reference:
a8aacccd79a3 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/sequence_format_converter commit d6ef80f9db43eae4f58b33f58b5ef6f8209907db
artbio
parents:
diff changeset
54 logging.info("But other nucleotides that A, T, G, C or N")
a8aacccd79a3 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/sequence_format_converter commit d6ef80f9db43eae4f58b33f58b5ef6f8209907db
artbio
parents:
diff changeset
55 sys.exit('input appears to be Fasta but with \
a8aacccd79a3 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/sequence_format_converter commit d6ef80f9db43eae4f58b33f58b5ef6f8209907db
artbio
parents:
diff changeset
56 unexpected nucleotides')
a8aacccd79a3 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/sequence_format_converter commit d6ef80f9db43eae4f58b33f58b5ef6f8209907db
artbio
parents:
diff changeset
57 format = 'fasta'
a8aacccd79a3 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/sequence_format_converter commit d6ef80f9db43eae4f58b33f58b5ef6f8209907db
artbio
parents:
diff changeset
58 elif line1[0] == '>' and line4[0] == '>':
a8aacccd79a3 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/sequence_format_converter commit d6ef80f9db43eae4f58b33f58b5ef6f8209907db
artbio
parents:
diff changeset
59 logging.info("'>' detected in lines 1 and 4")
a8aacccd79a3 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/sequence_format_converter commit d6ef80f9db43eae4f58b33f58b5ef6f8209907db
artbio
parents:
diff changeset
60 sequence = ''.join([line2, line3]).upper()
a8aacccd79a3 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/sequence_format_converter commit d6ef80f9db43eae4f58b33f58b5ef6f8209907db
artbio
parents:
diff changeset
61 nucleotides = set([base for base in sequence])
a8aacccd79a3 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/sequence_format_converter commit d6ef80f9db43eae4f58b33f58b5ef6f8209907db
artbio
parents:
diff changeset
62 for nucleotide in nucleotides:
a8aacccd79a3 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/sequence_format_converter commit d6ef80f9db43eae4f58b33f58b5ef6f8209907db
artbio
parents:
diff changeset
63 if nucleotide not in reference:
a8aacccd79a3 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/sequence_format_converter commit d6ef80f9db43eae4f58b33f58b5ef6f8209907db
artbio
parents:
diff changeset
64 logging.info("But other nucleotides that A, T, G, C or N")
a8aacccd79a3 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/sequence_format_converter commit d6ef80f9db43eae4f58b33f58b5ef6f8209907db
artbio
parents:
diff changeset
65 sys.exit('input appears to be Fasta but with \
a8aacccd79a3 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/sequence_format_converter commit d6ef80f9db43eae4f58b33f58b5ef6f8209907db
artbio
parents:
diff changeset
66 unexpected nucleotides')
a8aacccd79a3 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/sequence_format_converter commit d6ef80f9db43eae4f58b33f58b5ef6f8209907db
artbio
parents:
diff changeset
67 format = 'fasta'
a8aacccd79a3 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/sequence_format_converter commit d6ef80f9db43eae4f58b33f58b5ef6f8209907db
artbio
parents:
diff changeset
68 elif line1[0] == '>':
a8aacccd79a3 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/sequence_format_converter commit d6ef80f9db43eae4f58b33f58b5ef6f8209907db
artbio
parents:
diff changeset
69 logging.info("'>' detected in lines 1")
a8aacccd79a3 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/sequence_format_converter commit d6ef80f9db43eae4f58b33f58b5ef6f8209907db
artbio
parents:
diff changeset
70 sequence = ''.join([line2, line3, line4]).upper()
a8aacccd79a3 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/sequence_format_converter commit d6ef80f9db43eae4f58b33f58b5ef6f8209907db
artbio
parents:
diff changeset
71 nucleotides = set([base for base in sequence])
a8aacccd79a3 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/sequence_format_converter commit d6ef80f9db43eae4f58b33f58b5ef6f8209907db
artbio
parents:
diff changeset
72 for nucleotide in nucleotides:
a8aacccd79a3 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/sequence_format_converter commit d6ef80f9db43eae4f58b33f58b5ef6f8209907db
artbio
parents:
diff changeset
73 if nucleotide not in reference:
a8aacccd79a3 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/sequence_format_converter commit d6ef80f9db43eae4f58b33f58b5ef6f8209907db
artbio
parents:
diff changeset
74 logging.info("But other nucleotides that A, T, G, C or N")
a8aacccd79a3 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/sequence_format_converter commit d6ef80f9db43eae4f58b33f58b5ef6f8209907db
artbio
parents:
diff changeset
75 sys.exit('input appears to be Fasta but with \
a8aacccd79a3 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/sequence_format_converter commit d6ef80f9db43eae4f58b33f58b5ef6f8209907db
artbio
parents:
diff changeset
76 unexpected nucleotides')
a8aacccd79a3 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/sequence_format_converter commit d6ef80f9db43eae4f58b33f58b5ef6f8209907db
artbio
parents:
diff changeset
77 format = 'fasta'
a8aacccd79a3 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/sequence_format_converter commit d6ef80f9db43eae4f58b33f58b5ef6f8209907db
artbio
parents:
diff changeset
78 if format == 'fasta':
a8aacccd79a3 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/sequence_format_converter commit d6ef80f9db43eae4f58b33f58b5ef6f8209907db
artbio
parents:
diff changeset
79 try:
a8aacccd79a3 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/sequence_format_converter commit d6ef80f9db43eae4f58b33f58b5ef6f8209907db
artbio
parents:
diff changeset
80 for line in block:
a8aacccd79a3 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/sequence_format_converter commit d6ef80f9db43eae4f58b33f58b5ef6f8209907db
artbio
parents:
diff changeset
81 if line[0] == '>':
a8aacccd79a3 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/sequence_format_converter commit d6ef80f9db43eae4f58b33f58b5ef6f8209907db
artbio
parents:
diff changeset
82 int(line.split('_')[-1])
a8aacccd79a3 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/sequence_format_converter commit d6ef80f9db43eae4f58b33f58b5ef6f8209907db
artbio
parents:
diff changeset
83 return 'fastaw'
1
9ce7ccd468aa planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/sequence_format_converter commit dfecfb40f245a3cdb09dd1cfe37be4cb164ad2eb
artbio
parents: 0
diff changeset
84 except ValueError:
0
a8aacccd79a3 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/sequence_format_converter commit d6ef80f9db43eae4f58b33f58b5ef6f8209907db
artbio
parents:
diff changeset
85 return 'fasta'
a8aacccd79a3 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/sequence_format_converter commit d6ef80f9db43eae4f58b33f58b5ef6f8209907db
artbio
parents:
diff changeset
86 if line1[0] == '@' and line3[0] == '+':
a8aacccd79a3 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/sequence_format_converter commit d6ef80f9db43eae4f58b33f58b5ef6f8209907db
artbio
parents:
diff changeset
87 nucleotides = set([base for base in line2])
a8aacccd79a3 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/sequence_format_converter commit d6ef80f9db43eae4f58b33f58b5ef6f8209907db
artbio
parents:
diff changeset
88 for nucleotide in nucleotides:
a8aacccd79a3 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/sequence_format_converter commit d6ef80f9db43eae4f58b33f58b5ef6f8209907db
artbio
parents:
diff changeset
89 if nucleotide not in reference:
a8aacccd79a3 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/sequence_format_converter commit d6ef80f9db43eae4f58b33f58b5ef6f8209907db
artbio
parents:
diff changeset
90 logging.info("Looks like fastq input but other nucleotides \
a8aacccd79a3 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/sequence_format_converter commit d6ef80f9db43eae4f58b33f58b5ef6f8209907db
artbio
parents:
diff changeset
91 that A, T, G, C or N")
a8aacccd79a3 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/sequence_format_converter commit d6ef80f9db43eae4f58b33f58b5ef6f8209907db
artbio
parents:
diff changeset
92 sys.exit("input appears to be Fastq \
a8aacccd79a3 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/sequence_format_converter commit d6ef80f9db43eae4f58b33f58b5ef6f8209907db
artbio
parents:
diff changeset
93 but with unexpected nucleotides")
a8aacccd79a3 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/sequence_format_converter commit d6ef80f9db43eae4f58b33f58b5ef6f8209907db
artbio
parents:
diff changeset
94 return 'fastq'
a8aacccd79a3 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/sequence_format_converter commit d6ef80f9db43eae4f58b33f58b5ef6f8209907db
artbio
parents:
diff changeset
95 for line in block:
a8aacccd79a3 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/sequence_format_converter commit d6ef80f9db43eae4f58b33f58b5ef6f8209907db
artbio
parents:
diff changeset
96 if len(line.split('\t')) != 2:
a8aacccd79a3 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/sequence_format_converter commit d6ef80f9db43eae4f58b33f58b5ef6f8209907db
artbio
parents:
diff changeset
97 logging.info("No valid format detected")
a8aacccd79a3 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/sequence_format_converter commit d6ef80f9db43eae4f58b33f58b5ef6f8209907db
artbio
parents:
diff changeset
98 sys.exit('No valid format detected')
a8aacccd79a3 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/sequence_format_converter commit d6ef80f9db43eae4f58b33f58b5ef6f8209907db
artbio
parents:
diff changeset
99 try:
a8aacccd79a3 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/sequence_format_converter commit d6ef80f9db43eae4f58b33f58b5ef6f8209907db
artbio
parents:
diff changeset
100 int(line.split('\t')[-1])
1
9ce7ccd468aa planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/sequence_format_converter commit dfecfb40f245a3cdb09dd1cfe37be4cb164ad2eb
artbio
parents: 0
diff changeset
101 except ValueError:
0
a8aacccd79a3 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/sequence_format_converter commit d6ef80f9db43eae4f58b33f58b5ef6f8209907db
artbio
parents:
diff changeset
102 logging.info("No valid format detected")
a8aacccd79a3 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/sequence_format_converter commit d6ef80f9db43eae4f58b33f58b5ef6f8209907db
artbio
parents:
diff changeset
103 sys.exit('No valid format detected')
a8aacccd79a3 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/sequence_format_converter commit d6ef80f9db43eae4f58b33f58b5ef6f8209907db
artbio
parents:
diff changeset
104 for nucleotide in line.split('\t')[0]:
a8aacccd79a3 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/sequence_format_converter commit d6ef80f9db43eae4f58b33f58b5ef6f8209907db
artbio
parents:
diff changeset
105 if nucleotide not in reference:
a8aacccd79a3 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/sequence_format_converter commit d6ef80f9db43eae4f58b33f58b5ef6f8209907db
artbio
parents:
diff changeset
106 logging.info("No valid format detected")
a8aacccd79a3 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/sequence_format_converter commit d6ef80f9db43eae4f58b33f58b5ef6f8209907db
artbio
parents:
diff changeset
107 sys.exit('No valid format detected')
a8aacccd79a3 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/sequence_format_converter commit d6ef80f9db43eae4f58b33f58b5ef6f8209907db
artbio
parents:
diff changeset
108 return 'tabular'
a8aacccd79a3 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/sequence_format_converter commit d6ef80f9db43eae4f58b33f58b5ef6f8209907db
artbio
parents:
diff changeset
109
a8aacccd79a3 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/sequence_format_converter commit d6ef80f9db43eae4f58b33f58b5ef6f8209907db
artbio
parents:
diff changeset
110 def read(self, input, format):
a8aacccd79a3 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/sequence_format_converter commit d6ef80f9db43eae4f58b33f58b5ef6f8209907db
artbio
parents:
diff changeset
111 input = open(input, 'r')
a8aacccd79a3 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/sequence_format_converter commit d6ef80f9db43eae4f58b33f58b5ef6f8209907db
artbio
parents:
diff changeset
112 if format == 'fasta':
a8aacccd79a3 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/sequence_format_converter commit d6ef80f9db43eae4f58b33f58b5ef6f8209907db
artbio
parents:
diff changeset
113 try:
a8aacccd79a3 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/sequence_format_converter commit d6ef80f9db43eae4f58b33f58b5ef6f8209907db
artbio
parents:
diff changeset
114 self.readfasta(input)
1
9ce7ccd468aa planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/sequence_format_converter commit dfecfb40f245a3cdb09dd1cfe37be4cb164ad2eb
artbio
parents: 0
diff changeset
115 except Exception:
0
a8aacccd79a3 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/sequence_format_converter commit d6ef80f9db43eae4f58b33f58b5ef6f8209907db
artbio
parents:
diff changeset
116 logging.info("an error occured while reading fasta")
a8aacccd79a3 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/sequence_format_converter commit d6ef80f9db43eae4f58b33f58b5ef6f8209907db
artbio
parents:
diff changeset
117 elif format == 'fastaw':
a8aacccd79a3 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/sequence_format_converter commit d6ef80f9db43eae4f58b33f58b5ef6f8209907db
artbio
parents:
diff changeset
118 try:
a8aacccd79a3 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/sequence_format_converter commit d6ef80f9db43eae4f58b33f58b5ef6f8209907db
artbio
parents:
diff changeset
119 self.readfastaw(input)
1
9ce7ccd468aa planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/sequence_format_converter commit dfecfb40f245a3cdb09dd1cfe37be4cb164ad2eb
artbio
parents: 0
diff changeset
120 except Exception:
0
a8aacccd79a3 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/sequence_format_converter commit d6ef80f9db43eae4f58b33f58b5ef6f8209907db
artbio
parents:
diff changeset
121 logging.info("an error occured while reading fastaw")
a8aacccd79a3 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/sequence_format_converter commit d6ef80f9db43eae4f58b33f58b5ef6f8209907db
artbio
parents:
diff changeset
122 elif format == 'tabular':
a8aacccd79a3 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/sequence_format_converter commit d6ef80f9db43eae4f58b33f58b5ef6f8209907db
artbio
parents:
diff changeset
123 try:
a8aacccd79a3 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/sequence_format_converter commit d6ef80f9db43eae4f58b33f58b5ef6f8209907db
artbio
parents:
diff changeset
124 self.readtabular(input)
1
9ce7ccd468aa planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/sequence_format_converter commit dfecfb40f245a3cdb09dd1cfe37be4cb164ad2eb
artbio
parents: 0
diff changeset
125 except Exception:
0
a8aacccd79a3 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/sequence_format_converter commit d6ef80f9db43eae4f58b33f58b5ef6f8209907db
artbio
parents:
diff changeset
126 logging.info("an error occured while reading tabular")
a8aacccd79a3 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/sequence_format_converter commit d6ef80f9db43eae4f58b33f58b5ef6f8209907db
artbio
parents:
diff changeset
127 elif format == 'fastq':
a8aacccd79a3 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/sequence_format_converter commit d6ef80f9db43eae4f58b33f58b5ef6f8209907db
artbio
parents:
diff changeset
128 try:
a8aacccd79a3 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/sequence_format_converter commit d6ef80f9db43eae4f58b33f58b5ef6f8209907db
artbio
parents:
diff changeset
129 self.readfastq(input)
1
9ce7ccd468aa planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/sequence_format_converter commit dfecfb40f245a3cdb09dd1cfe37be4cb164ad2eb
artbio
parents: 0
diff changeset
130 except Exception:
0
a8aacccd79a3 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/sequence_format_converter commit d6ef80f9db43eae4f58b33f58b5ef6f8209907db
artbio
parents:
diff changeset
131 logging.info("an error occured while reading fastq")
a8aacccd79a3 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/sequence_format_converter commit d6ef80f9db43eae4f58b33f58b5ef6f8209907db
artbio
parents:
diff changeset
132 else:
a8aacccd79a3 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/sequence_format_converter commit d6ef80f9db43eae4f58b33f58b5ef6f8209907db
artbio
parents:
diff changeset
133 logging.info("no valid format detected")
a8aacccd79a3 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/sequence_format_converter commit d6ef80f9db43eae4f58b33f58b5ef6f8209907db
artbio
parents:
diff changeset
134 sys.exit('No valid format detected')
a8aacccd79a3 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/sequence_format_converter commit d6ef80f9db43eae4f58b33f58b5ef6f8209907db
artbio
parents:
diff changeset
135
a8aacccd79a3 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/sequence_format_converter commit d6ef80f9db43eae4f58b33f58b5ef6f8209907db
artbio
parents:
diff changeset
136 def readfastaw(self, input):
a8aacccd79a3 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/sequence_format_converter commit d6ef80f9db43eae4f58b33f58b5ef6f8209907db
artbio
parents:
diff changeset
137 for line in input:
a8aacccd79a3 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/sequence_format_converter commit d6ef80f9db43eae4f58b33f58b5ef6f8209907db
artbio
parents:
diff changeset
138 if line[0] == ">":
a8aacccd79a3 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/sequence_format_converter commit d6ef80f9db43eae4f58b33f58b5ef6f8209907db
artbio
parents:
diff changeset
139 weigth = int(line[:-1].split("_")[-1])
a8aacccd79a3 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/sequence_format_converter commit d6ef80f9db43eae4f58b33f58b5ef6f8209907db
artbio
parents:
diff changeset
140 else:
a8aacccd79a3 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/sequence_format_converter commit d6ef80f9db43eae4f58b33f58b5ef6f8209907db
artbio
parents:
diff changeset
141 self.seqdic[line[:-1]] += weigth
a8aacccd79a3 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/sequence_format_converter commit d6ef80f9db43eae4f58b33f58b5ef6f8209907db
artbio
parents:
diff changeset
142 input.close()
a8aacccd79a3 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/sequence_format_converter commit d6ef80f9db43eae4f58b33f58b5ef6f8209907db
artbio
parents:
diff changeset
143
a8aacccd79a3 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/sequence_format_converter commit d6ef80f9db43eae4f58b33f58b5ef6f8209907db
artbio
parents:
diff changeset
144 def readfasta(self, input):
a8aacccd79a3 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/sequence_format_converter commit d6ef80f9db43eae4f58b33f58b5ef6f8209907db
artbio
parents:
diff changeset
145 ''' this method is able to read multi-line fasta sequence'''
a8aacccd79a3 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/sequence_format_converter commit d6ef80f9db43eae4f58b33f58b5ef6f8209907db
artbio
parents:
diff changeset
146 for line in input:
a8aacccd79a3 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/sequence_format_converter commit d6ef80f9db43eae4f58b33f58b5ef6f8209907db
artbio
parents:
diff changeset
147 if line[0] == ">":
a8aacccd79a3 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/sequence_format_converter commit d6ef80f9db43eae4f58b33f58b5ef6f8209907db
artbio
parents:
diff changeset
148 try:
a8aacccd79a3 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/sequence_format_converter commit d6ef80f9db43eae4f58b33f58b5ef6f8209907db
artbio
parents:
diff changeset
149 # to dump the sequence of the previous item
a8aacccd79a3 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/sequence_format_converter commit d6ef80f9db43eae4f58b33f58b5ef6f8209907db
artbio
parents:
diff changeset
150 # try because of first missing stringlist variable
a8aacccd79a3 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/sequence_format_converter commit d6ef80f9db43eae4f58b33f58b5ef6f8209907db
artbio
parents:
diff changeset
151 self.seqdic["".join(stringlist)] += 1
a8aacccd79a3 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/sequence_format_converter commit d6ef80f9db43eae4f58b33f58b5ef6f8209907db
artbio
parents:
diff changeset
152 except NameError:
a8aacccd79a3 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/sequence_format_converter commit d6ef80f9db43eae4f58b33f58b5ef6f8209907db
artbio
parents:
diff changeset
153 pass
a8aacccd79a3 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/sequence_format_converter commit d6ef80f9db43eae4f58b33f58b5ef6f8209907db
artbio
parents:
diff changeset
154 stringlist = []
a8aacccd79a3 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/sequence_format_converter commit d6ef80f9db43eae4f58b33f58b5ef6f8209907db
artbio
parents:
diff changeset
155 else:
a8aacccd79a3 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/sequence_format_converter commit d6ef80f9db43eae4f58b33f58b5ef6f8209907db
artbio
parents:
diff changeset
156 try:
a8aacccd79a3 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/sequence_format_converter commit d6ef80f9db43eae4f58b33f58b5ef6f8209907db
artbio
parents:
diff changeset
157 stringlist.append(line[:-1])
a8aacccd79a3 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/sequence_format_converter commit d6ef80f9db43eae4f58b33f58b5ef6f8209907db
artbio
parents:
diff changeset
158 except UnboundLocalError:
a8aacccd79a3 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/sequence_format_converter commit d6ef80f9db43eae4f58b33f58b5ef6f8209907db
artbio
parents:
diff changeset
159 # if file went through filter and contains only empty lines
a8aacccd79a3 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/sequence_format_converter commit d6ef80f9db43eae4f58b33f58b5ef6f8209907db
artbio
parents:
diff changeset
160 logging.info("first line is empty.")
a8aacccd79a3 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/sequence_format_converter commit d6ef80f9db43eae4f58b33f58b5ef6f8209907db
artbio
parents:
diff changeset
161 try:
a8aacccd79a3 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/sequence_format_converter commit d6ef80f9db43eae4f58b33f58b5ef6f8209907db
artbio
parents:
diff changeset
162 self.seqdic["".join(stringlist)] += 1 # for the last sequence
a8aacccd79a3 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/sequence_format_converter commit d6ef80f9db43eae4f58b33f58b5ef6f8209907db
artbio
parents:
diff changeset
163 except NameError:
a8aacccd79a3 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/sequence_format_converter commit d6ef80f9db43eae4f58b33f58b5ef6f8209907db
artbio
parents:
diff changeset
164 logging.info("input file has not fasta sequences.")
a8aacccd79a3 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/sequence_format_converter commit d6ef80f9db43eae4f58b33f58b5ef6f8209907db
artbio
parents:
diff changeset
165 input.close()
a8aacccd79a3 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/sequence_format_converter commit d6ef80f9db43eae4f58b33f58b5ef6f8209907db
artbio
parents:
diff changeset
166
a8aacccd79a3 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/sequence_format_converter commit d6ef80f9db43eae4f58b33f58b5ef6f8209907db
artbio
parents:
diff changeset
167 def readtabular(self, input):
a8aacccd79a3 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/sequence_format_converter commit d6ef80f9db43eae4f58b33f58b5ef6f8209907db
artbio
parents:
diff changeset
168 for line in input:
a8aacccd79a3 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/sequence_format_converter commit d6ef80f9db43eae4f58b33f58b5ef6f8209907db
artbio
parents:
diff changeset
169 fields = line[:-1].split('\t')
a8aacccd79a3 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/sequence_format_converter commit d6ef80f9db43eae4f58b33f58b5ef6f8209907db
artbio
parents:
diff changeset
170 self.seqdic[fields[0]] += int(fields[1])
a8aacccd79a3 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/sequence_format_converter commit d6ef80f9db43eae4f58b33f58b5ef6f8209907db
artbio
parents:
diff changeset
171 input.close()
a8aacccd79a3 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/sequence_format_converter commit d6ef80f9db43eae4f58b33f58b5ef6f8209907db
artbio
parents:
diff changeset
172
a8aacccd79a3 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/sequence_format_converter commit d6ef80f9db43eae4f58b33f58b5ef6f8209907db
artbio
parents:
diff changeset
173 def readfastq(self, input):
a8aacccd79a3 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/sequence_format_converter commit d6ef80f9db43eae4f58b33f58b5ef6f8209907db
artbio
parents:
diff changeset
174 linecount = 0
a8aacccd79a3 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/sequence_format_converter commit d6ef80f9db43eae4f58b33f58b5ef6f8209907db
artbio
parents:
diff changeset
175 for line in input:
a8aacccd79a3 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/sequence_format_converter commit d6ef80f9db43eae4f58b33f58b5ef6f8209907db
artbio
parents:
diff changeset
176 linecount += 1
a8aacccd79a3 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/sequence_format_converter commit d6ef80f9db43eae4f58b33f58b5ef6f8209907db
artbio
parents:
diff changeset
177 if linecount % 4 == 2:
a8aacccd79a3 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/sequence_format_converter commit d6ef80f9db43eae4f58b33f58b5ef6f8209907db
artbio
parents:
diff changeset
178 self.seqdic[line[:-1]] += 1
a8aacccd79a3 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/sequence_format_converter commit d6ef80f9db43eae4f58b33f58b5ef6f8209907db
artbio
parents:
diff changeset
179 input.close()
a8aacccd79a3 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/sequence_format_converter commit d6ef80f9db43eae4f58b33f58b5ef6f8209907db
artbio
parents:
diff changeset
180
a8aacccd79a3 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/sequence_format_converter commit d6ef80f9db43eae4f58b33f58b5ef6f8209907db
artbio
parents:
diff changeset
181 def write(self, output, format='fasta'):
a8aacccd79a3 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/sequence_format_converter commit d6ef80f9db43eae4f58b33f58b5ef6f8209907db
artbio
parents:
diff changeset
182 if format == 'fasta':
a8aacccd79a3 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/sequence_format_converter commit d6ef80f9db43eae4f58b33f58b5ef6f8209907db
artbio
parents:
diff changeset
183 headercount = 0
a8aacccd79a3 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/sequence_format_converter commit d6ef80f9db43eae4f58b33f58b5ef6f8209907db
artbio
parents:
diff changeset
184 for seq in sorted(self.seqdic, key=self.seqdic.get, reverse=True):
a8aacccd79a3 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/sequence_format_converter commit d6ef80f9db43eae4f58b33f58b5ef6f8209907db
artbio
parents:
diff changeset
185 for i in range(self.seqdic[seq]):
a8aacccd79a3 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/sequence_format_converter commit d6ef80f9db43eae4f58b33f58b5ef6f8209907db
artbio
parents:
diff changeset
186 headercount += 1
a8aacccd79a3 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/sequence_format_converter commit d6ef80f9db43eae4f58b33f58b5ef6f8209907db
artbio
parents:
diff changeset
187 output.write('>%s\n%s\n' % (headercount, seq))
a8aacccd79a3 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/sequence_format_converter commit d6ef80f9db43eae4f58b33f58b5ef6f8209907db
artbio
parents:
diff changeset
188 elif format == 'fastaw':
a8aacccd79a3 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/sequence_format_converter commit d6ef80f9db43eae4f58b33f58b5ef6f8209907db
artbio
parents:
diff changeset
189 headercount = 0
a8aacccd79a3 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/sequence_format_converter commit d6ef80f9db43eae4f58b33f58b5ef6f8209907db
artbio
parents:
diff changeset
190 for seq in sorted(self.seqdic, key=self.seqdic.get, reverse=True):
a8aacccd79a3 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/sequence_format_converter commit d6ef80f9db43eae4f58b33f58b5ef6f8209907db
artbio
parents:
diff changeset
191 headercount += 1
a8aacccd79a3 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/sequence_format_converter commit d6ef80f9db43eae4f58b33f58b5ef6f8209907db
artbio
parents:
diff changeset
192 output.write('>%s_%s\n%s\n' % (headercount,
a8aacccd79a3 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/sequence_format_converter commit d6ef80f9db43eae4f58b33f58b5ef6f8209907db
artbio
parents:
diff changeset
193 self.seqdic[seq], seq))
a8aacccd79a3 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/sequence_format_converter commit d6ef80f9db43eae4f58b33f58b5ef6f8209907db
artbio
parents:
diff changeset
194 elif format == 'tabular':
a8aacccd79a3 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/sequence_format_converter commit d6ef80f9db43eae4f58b33f58b5ef6f8209907db
artbio
parents:
diff changeset
195 for seq in sorted(self.seqdic, key=self.seqdic.get, reverse=True):
a8aacccd79a3 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/sequence_format_converter commit d6ef80f9db43eae4f58b33f58b5ef6f8209907db
artbio
parents:
diff changeset
196 output.write('%s\t%s\n' % (seq, self.seqdic[seq]))
a8aacccd79a3 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/sequence_format_converter commit d6ef80f9db43eae4f58b33f58b5ef6f8209907db
artbio
parents:
diff changeset
197 output.close()
a8aacccd79a3 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/sequence_format_converter commit d6ef80f9db43eae4f58b33f58b5ef6f8209907db
artbio
parents:
diff changeset
198
a8aacccd79a3 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/sequence_format_converter commit d6ef80f9db43eae4f58b33f58b5ef6f8209907db
artbio
parents:
diff changeset
199
a8aacccd79a3 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/sequence_format_converter commit d6ef80f9db43eae4f58b33f58b5ef6f8209907db
artbio
parents:
diff changeset
200 def main(input, output, format):
a8aacccd79a3 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/sequence_format_converter commit d6ef80f9db43eae4f58b33f58b5ef6f8209907db
artbio
parents:
diff changeset
201 Sequencing(input, output, format)
a8aacccd79a3 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/sequence_format_converter commit d6ef80f9db43eae4f58b33f58b5ef6f8209907db
artbio
parents:
diff changeset
202
a8aacccd79a3 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/sequence_format_converter commit d6ef80f9db43eae4f58b33f58b5ef6f8209907db
artbio
parents:
diff changeset
203
a8aacccd79a3 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/sequence_format_converter commit d6ef80f9db43eae4f58b33f58b5ef6f8209907db
artbio
parents:
diff changeset
204 if __name__ == "__main__":
a8aacccd79a3 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/sequence_format_converter commit d6ef80f9db43eae4f58b33f58b5ef6f8209907db
artbio
parents:
diff changeset
205 args = Parser()
a8aacccd79a3 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/sequence_format_converter commit d6ef80f9db43eae4f58b33f58b5ef6f8209907db
artbio
parents:
diff changeset
206 log = logging.getLogger(__name__)
a8aacccd79a3 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/sequence_format_converter commit d6ef80f9db43eae4f58b33f58b5ef6f8209907db
artbio
parents:
diff changeset
207 logging.basicConfig(stream=sys.stdout, level=logging.INFO)
a8aacccd79a3 planemo upload for repository https://github.com/ARTbio/tools-artbio/tree/master/tools/sequence_format_converter commit d6ef80f9db43eae4f58b33f58b5ef6f8209907db
artbio
parents:
diff changeset
208 main(args.input, args.output, args.format)