Mercurial > repos > nick > duplex
annotate utils/get_msa.py @ 4:af383638de66 draft
planemo upload commit 022984f323d3da44f70b3bf79c684cfd8dda3f61-dirty
author | nick |
---|---|
date | Mon, 23 Nov 2015 18:44:23 -0500 |
parents | |
children |
rev | line source |
---|---|
4
af383638de66
planemo upload commit 022984f323d3da44f70b3bf79c684cfd8dda3f61-dirty
nick
parents:
diff
changeset
|
1 #!/usr/bin/env python |
af383638de66
planemo upload commit 022984f323d3da44f70b3bf79c684cfd8dda3f61-dirty
nick
parents:
diff
changeset
|
2 from __future__ import division |
af383638de66
planemo upload commit 022984f323d3da44f70b3bf79c684cfd8dda3f61-dirty
nick
parents:
diff
changeset
|
3 import os |
af383638de66
planemo upload commit 022984f323d3da44f70b3bf79c684cfd8dda3f61-dirty
nick
parents:
diff
changeset
|
4 import sys |
af383638de66
planemo upload commit 022984f323d3da44f70b3bf79c684cfd8dda3f61-dirty
nick
parents:
diff
changeset
|
5 import argparse |
af383638de66
planemo upload commit 022984f323d3da44f70b3bf79c684cfd8dda3f61-dirty
nick
parents:
diff
changeset
|
6 import tempfile |
af383638de66
planemo upload commit 022984f323d3da44f70b3bf79c684cfd8dda3f61-dirty
nick
parents:
diff
changeset
|
7 import subprocess |
af383638de66
planemo upload commit 022984f323d3da44f70b3bf79c684cfd8dda3f61-dirty
nick
parents:
diff
changeset
|
8 sys.path.append(os.path.dirname(os.path.dirname(os.path.realpath(__file__)))) |
af383638de66
planemo upload commit 022984f323d3da44f70b3bf79c684cfd8dda3f61-dirty
nick
parents:
diff
changeset
|
9 import consensus |
af383638de66
planemo upload commit 022984f323d3da44f70b3bf79c684cfd8dda3f61-dirty
nick
parents:
diff
changeset
|
10 import seqtools |
af383638de66
planemo upload commit 022984f323d3da44f70b3bf79c684cfd8dda3f61-dirty
nick
parents:
diff
changeset
|
11 |
af383638de66
planemo upload commit 022984f323d3da44f70b3bf79c684cfd8dda3f61-dirty
nick
parents:
diff
changeset
|
12 OPT_DEFAULTS = {'format':'plain', 'qual':20, 'qual_format':'sanger'} |
af383638de66
planemo upload commit 022984f323d3da44f70b3bf79c684cfd8dda3f61-dirty
nick
parents:
diff
changeset
|
13 USAGE = "%(prog)s [options]" |
af383638de66
planemo upload commit 022984f323d3da44f70b3bf79c684cfd8dda3f61-dirty
nick
parents:
diff
changeset
|
14 DESCRIPTION = """""" |
af383638de66
planemo upload commit 022984f323d3da44f70b3bf79c684cfd8dda3f61-dirty
nick
parents:
diff
changeset
|
15 |
af383638de66
planemo upload commit 022984f323d3da44f70b3bf79c684cfd8dda3f61-dirty
nick
parents:
diff
changeset
|
16 |
af383638de66
planemo upload commit 022984f323d3da44f70b3bf79c684cfd8dda3f61-dirty
nick
parents:
diff
changeset
|
17 def main(argv): |
af383638de66
planemo upload commit 022984f323d3da44f70b3bf79c684cfd8dda3f61-dirty
nick
parents:
diff
changeset
|
18 |
af383638de66
planemo upload commit 022984f323d3da44f70b3bf79c684cfd8dda3f61-dirty
nick
parents:
diff
changeset
|
19 parser = argparse.ArgumentParser(description=DESCRIPTION) |
af383638de66
planemo upload commit 022984f323d3da44f70b3bf79c684cfd8dda3f61-dirty
nick
parents:
diff
changeset
|
20 parser.set_defaults(**OPT_DEFAULTS) |
af383638de66
planemo upload commit 022984f323d3da44f70b3bf79c684cfd8dda3f61-dirty
nick
parents:
diff
changeset
|
21 |
af383638de66
planemo upload commit 022984f323d3da44f70b3bf79c684cfd8dda3f61-dirty
nick
parents:
diff
changeset
|
22 parser.add_argument('seqs', metavar='sequence', nargs='*', |
af383638de66
planemo upload commit 022984f323d3da44f70b3bf79c684cfd8dda3f61-dirty
nick
parents:
diff
changeset
|
23 help='The alignment.') |
af383638de66
planemo upload commit 022984f323d3da44f70b3bf79c684cfd8dda3f61-dirty
nick
parents:
diff
changeset
|
24 parser.add_argument('-i', '--input', |
af383638de66
planemo upload commit 022984f323d3da44f70b3bf79c684cfd8dda3f61-dirty
nick
parents:
diff
changeset
|
25 help='Provide the sequences in this input file instead of as command-line arguments. ' |
af383638de66
planemo upload commit 022984f323d3da44f70b3bf79c684cfd8dda3f61-dirty
nick
parents:
diff
changeset
|
26 'Give "-" to use stdin.') |
af383638de66
planemo upload commit 022984f323d3da44f70b3bf79c684cfd8dda3f61-dirty
nick
parents:
diff
changeset
|
27 parser.add_argument('-f', '--format', choices=('plain', 'duplex'), |
af383638de66
planemo upload commit 022984f323d3da44f70b3bf79c684cfd8dda3f61-dirty
nick
parents:
diff
changeset
|
28 help='Input format. "plain" is a simple list of the sequences, one on each line. "duplex" is ' |
af383638de66
planemo upload commit 022984f323d3da44f70b3bf79c684cfd8dda3f61-dirty
nick
parents:
diff
changeset
|
29 'the 8-column format of the family-sorted read data from the duplex pipeline. It must be ' |
af383638de66
planemo upload commit 022984f323d3da44f70b3bf79c684cfd8dda3f61-dirty
nick
parents:
diff
changeset
|
30 'the read pairs from a single alpha/beta barcode combination (both the alpha-beta and ' |
af383638de66
planemo upload commit 022984f323d3da44f70b3bf79c684cfd8dda3f61-dirty
nick
parents:
diff
changeset
|
31 'beta-alpha strands). If "duplex" is given, you must also specify which of the four ' |
af383638de66
planemo upload commit 022984f323d3da44f70b3bf79c684cfd8dda3f61-dirty
nick
parents:
diff
changeset
|
32 'possible alignments to output with --mate and --order.') |
af383638de66
planemo upload commit 022984f323d3da44f70b3bf79c684cfd8dda3f61-dirty
nick
parents:
diff
changeset
|
33 parser.add_argument('-m', '--mate', type=int, choices=(1, 2)) |
af383638de66
planemo upload commit 022984f323d3da44f70b3bf79c684cfd8dda3f61-dirty
nick
parents:
diff
changeset
|
34 parser.add_argument('-o', '--order', choices=('ab', 'ba')) |
af383638de66
planemo upload commit 022984f323d3da44f70b3bf79c684cfd8dda3f61-dirty
nick
parents:
diff
changeset
|
35 parser.add_argument('-F', '--qual-format', choices=('sanger',)) |
af383638de66
planemo upload commit 022984f323d3da44f70b3bf79c684cfd8dda3f61-dirty
nick
parents:
diff
changeset
|
36 parser.add_argument('-q', '--qual', type=int, |
af383638de66
planemo upload commit 022984f323d3da44f70b3bf79c684cfd8dda3f61-dirty
nick
parents:
diff
changeset
|
37 help='Quality threshold: Default: %(default)s') |
af383638de66
planemo upload commit 022984f323d3da44f70b3bf79c684cfd8dda3f61-dirty
nick
parents:
diff
changeset
|
38 |
af383638de66
planemo upload commit 022984f323d3da44f70b3bf79c684cfd8dda3f61-dirty
nick
parents:
diff
changeset
|
39 args = parser.parse_args(argv[1:]) |
af383638de66
planemo upload commit 022984f323d3da44f70b3bf79c684cfd8dda3f61-dirty
nick
parents:
diff
changeset
|
40 |
af383638de66
planemo upload commit 022984f323d3da44f70b3bf79c684cfd8dda3f61-dirty
nick
parents:
diff
changeset
|
41 qual_thres = ' ' |
af383638de66
planemo upload commit 022984f323d3da44f70b3bf79c684cfd8dda3f61-dirty
nick
parents:
diff
changeset
|
42 if args.qual_format == 'sanger': |
af383638de66
planemo upload commit 022984f323d3da44f70b3bf79c684cfd8dda3f61-dirty
nick
parents:
diff
changeset
|
43 qual_thres = chr(args.qual + 33) |
af383638de66
planemo upload commit 022984f323d3da44f70b3bf79c684cfd8dda3f61-dirty
nick
parents:
diff
changeset
|
44 else: |
af383638de66
planemo upload commit 022984f323d3da44f70b3bf79c684cfd8dda3f61-dirty
nick
parents:
diff
changeset
|
45 fail('Error: Unsupported FASTQ quality format "{}".'.format(args.qual_format)) |
af383638de66
planemo upload commit 022984f323d3da44f70b3bf79c684cfd8dda3f61-dirty
nick
parents:
diff
changeset
|
46 # Check arguments. |
af383638de66
planemo upload commit 022984f323d3da44f70b3bf79c684cfd8dda3f61-dirty
nick
parents:
diff
changeset
|
47 if not (args.seqs or args.input): |
af383638de66
planemo upload commit 022984f323d3da44f70b3bf79c684cfd8dda3f61-dirty
nick
parents:
diff
changeset
|
48 fail('Error: You must provide sequences either in a file with --input or as arguments.') |
af383638de66
planemo upload commit 022984f323d3da44f70b3bf79c684cfd8dda3f61-dirty
nick
parents:
diff
changeset
|
49 elif args.seqs and args.input: |
af383638de66
planemo upload commit 022984f323d3da44f70b3bf79c684cfd8dda3f61-dirty
nick
parents:
diff
changeset
|
50 fail('Error: You cannot provide sequences in both a file and command-line arguments.') |
af383638de66
planemo upload commit 022984f323d3da44f70b3bf79c684cfd8dda3f61-dirty
nick
parents:
diff
changeset
|
51 if args.format == 'duplex' and not (args.mate and args.order): |
af383638de66
planemo upload commit 022984f323d3da44f70b3bf79c684cfd8dda3f61-dirty
nick
parents:
diff
changeset
|
52 fail('Error: If the --format is duplex, you must specify a --mate and --order.') |
af383638de66
planemo upload commit 022984f323d3da44f70b3bf79c684cfd8dda3f61-dirty
nick
parents:
diff
changeset
|
53 |
af383638de66
planemo upload commit 022984f323d3da44f70b3bf79c684cfd8dda3f61-dirty
nick
parents:
diff
changeset
|
54 # Read input. |
af383638de66
planemo upload commit 022984f323d3da44f70b3bf79c684cfd8dda3f61-dirty
nick
parents:
diff
changeset
|
55 quals = [] |
af383638de66
planemo upload commit 022984f323d3da44f70b3bf79c684cfd8dda3f61-dirty
nick
parents:
diff
changeset
|
56 if args.input: |
af383638de66
planemo upload commit 022984f323d3da44f70b3bf79c684cfd8dda3f61-dirty
nick
parents:
diff
changeset
|
57 if args.format == 'plain': |
af383638de66
planemo upload commit 022984f323d3da44f70b3bf79c684cfd8dda3f61-dirty
nick
parents:
diff
changeset
|
58 if args.input == '-': |
af383638de66
planemo upload commit 022984f323d3da44f70b3bf79c684cfd8dda3f61-dirty
nick
parents:
diff
changeset
|
59 seqs = [line.strip() for line in sys.stdin] |
af383638de66
planemo upload commit 022984f323d3da44f70b3bf79c684cfd8dda3f61-dirty
nick
parents:
diff
changeset
|
60 else: |
af383638de66
planemo upload commit 022984f323d3da44f70b3bf79c684cfd8dda3f61-dirty
nick
parents:
diff
changeset
|
61 with open(args.input) as infile: |
af383638de66
planemo upload commit 022984f323d3da44f70b3bf79c684cfd8dda3f61-dirty
nick
parents:
diff
changeset
|
62 seqs = [line.strip() for line in infile] |
af383638de66
planemo upload commit 022984f323d3da44f70b3bf79c684cfd8dda3f61-dirty
nick
parents:
diff
changeset
|
63 elif args.format == 'duplex': |
af383638de66
planemo upload commit 022984f323d3da44f70b3bf79c684cfd8dda3f61-dirty
nick
parents:
diff
changeset
|
64 if args.input == '-': |
af383638de66
planemo upload commit 022984f323d3da44f70b3bf79c684cfd8dda3f61-dirty
nick
parents:
diff
changeset
|
65 (seqs, quals) = parse_duplex(sys.stdin, args.mate, args.order) |
af383638de66
planemo upload commit 022984f323d3da44f70b3bf79c684cfd8dda3f61-dirty
nick
parents:
diff
changeset
|
66 else: |
af383638de66
planemo upload commit 022984f323d3da44f70b3bf79c684cfd8dda3f61-dirty
nick
parents:
diff
changeset
|
67 with open(args.input) as infile: |
af383638de66
planemo upload commit 022984f323d3da44f70b3bf79c684cfd8dda3f61-dirty
nick
parents:
diff
changeset
|
68 (seqs, quals) = parse_duplex(infile, args.mate, args.order) |
af383638de66
planemo upload commit 022984f323d3da44f70b3bf79c684cfd8dda3f61-dirty
nick
parents:
diff
changeset
|
69 else: |
af383638de66
planemo upload commit 022984f323d3da44f70b3bf79c684cfd8dda3f61-dirty
nick
parents:
diff
changeset
|
70 seqs = args.seqs |
af383638de66
planemo upload commit 022984f323d3da44f70b3bf79c684cfd8dda3f61-dirty
nick
parents:
diff
changeset
|
71 |
af383638de66
planemo upload commit 022984f323d3da44f70b3bf79c684cfd8dda3f61-dirty
nick
parents:
diff
changeset
|
72 align = make_msa(seqs) |
af383638de66
planemo upload commit 022984f323d3da44f70b3bf79c684cfd8dda3f61-dirty
nick
parents:
diff
changeset
|
73 if quals: |
af383638de66
planemo upload commit 022984f323d3da44f70b3bf79c684cfd8dda3f61-dirty
nick
parents:
diff
changeset
|
74 quals = seqtools.transfer_gaps_multi(quals, align, gap_char_out=' ') |
af383638de66
planemo upload commit 022984f323d3da44f70b3bf79c684cfd8dda3f61-dirty
nick
parents:
diff
changeset
|
75 cons = consensus.get_consensus(align, quals, qual_thres=qual_thres, gapped=True) |
af383638de66
planemo upload commit 022984f323d3da44f70b3bf79c684cfd8dda3f61-dirty
nick
parents:
diff
changeset
|
76 |
af383638de66
planemo upload commit 022984f323d3da44f70b3bf79c684cfd8dda3f61-dirty
nick
parents:
diff
changeset
|
77 output = format_alignment(cons, align, quals, qual_thres=ord(qual_thres)) |
af383638de66
planemo upload commit 022984f323d3da44f70b3bf79c684cfd8dda3f61-dirty
nick
parents:
diff
changeset
|
78 |
af383638de66
planemo upload commit 022984f323d3da44f70b3bf79c684cfd8dda3f61-dirty
nick
parents:
diff
changeset
|
79 for seq in output: |
af383638de66
planemo upload commit 022984f323d3da44f70b3bf79c684cfd8dda3f61-dirty
nick
parents:
diff
changeset
|
80 print seq |
af383638de66
planemo upload commit 022984f323d3da44f70b3bf79c684cfd8dda3f61-dirty
nick
parents:
diff
changeset
|
81 |
af383638de66
planemo upload commit 022984f323d3da44f70b3bf79c684cfd8dda3f61-dirty
nick
parents:
diff
changeset
|
82 |
af383638de66
planemo upload commit 022984f323d3da44f70b3bf79c684cfd8dda3f61-dirty
nick
parents:
diff
changeset
|
83 def parse_duplex(infile, mate, order): |
af383638de66
planemo upload commit 022984f323d3da44f70b3bf79c684cfd8dda3f61-dirty
nick
parents:
diff
changeset
|
84 seqs = [] |
af383638de66
planemo upload commit 022984f323d3da44f70b3bf79c684cfd8dda3f61-dirty
nick
parents:
diff
changeset
|
85 quals = [] |
af383638de66
planemo upload commit 022984f323d3da44f70b3bf79c684cfd8dda3f61-dirty
nick
parents:
diff
changeset
|
86 for line in infile: |
af383638de66
planemo upload commit 022984f323d3da44f70b3bf79c684cfd8dda3f61-dirty
nick
parents:
diff
changeset
|
87 (bar, this_order, name1, seq1, qual1, name2, seq2, qual2) = line.rstrip('\r\n').split('\t') |
af383638de66
planemo upload commit 022984f323d3da44f70b3bf79c684cfd8dda3f61-dirty
nick
parents:
diff
changeset
|
88 if this_order == order: |
af383638de66
planemo upload commit 022984f323d3da44f70b3bf79c684cfd8dda3f61-dirty
nick
parents:
diff
changeset
|
89 if mate == 1: |
af383638de66
planemo upload commit 022984f323d3da44f70b3bf79c684cfd8dda3f61-dirty
nick
parents:
diff
changeset
|
90 seqs.append(seq1) |
af383638de66
planemo upload commit 022984f323d3da44f70b3bf79c684cfd8dda3f61-dirty
nick
parents:
diff
changeset
|
91 quals.append(qual1) |
af383638de66
planemo upload commit 022984f323d3da44f70b3bf79c684cfd8dda3f61-dirty
nick
parents:
diff
changeset
|
92 elif mate == 2: |
af383638de66
planemo upload commit 022984f323d3da44f70b3bf79c684cfd8dda3f61-dirty
nick
parents:
diff
changeset
|
93 seqs.append(seq2) |
af383638de66
planemo upload commit 022984f323d3da44f70b3bf79c684cfd8dda3f61-dirty
nick
parents:
diff
changeset
|
94 quals.append(qual2) |
af383638de66
planemo upload commit 022984f323d3da44f70b3bf79c684cfd8dda3f61-dirty
nick
parents:
diff
changeset
|
95 return seqs, quals |
af383638de66
planemo upload commit 022984f323d3da44f70b3bf79c684cfd8dda3f61-dirty
nick
parents:
diff
changeset
|
96 |
af383638de66
planemo upload commit 022984f323d3da44f70b3bf79c684cfd8dda3f61-dirty
nick
parents:
diff
changeset
|
97 |
af383638de66
planemo upload commit 022984f323d3da44f70b3bf79c684cfd8dda3f61-dirty
nick
parents:
diff
changeset
|
98 def make_msa(seqs): |
af383638de66
planemo upload commit 022984f323d3da44f70b3bf79c684cfd8dda3f61-dirty
nick
parents:
diff
changeset
|
99 """Perform a multiple sequence alignment on a set of sequences. |
af383638de66
planemo upload commit 022984f323d3da44f70b3bf79c684cfd8dda3f61-dirty
nick
parents:
diff
changeset
|
100 Uses MAFFT.""" |
af383638de66
planemo upload commit 022984f323d3da44f70b3bf79c684cfd8dda3f61-dirty
nick
parents:
diff
changeset
|
101 i = 0 |
af383638de66
planemo upload commit 022984f323d3da44f70b3bf79c684cfd8dda3f61-dirty
nick
parents:
diff
changeset
|
102 #TODO: Replace with tempfile.mkstemp()? |
af383638de66
planemo upload commit 022984f323d3da44f70b3bf79c684cfd8dda3f61-dirty
nick
parents:
diff
changeset
|
103 with tempfile.NamedTemporaryFile('w', delete=False, prefix='msa.') as family_file: |
af383638de66
planemo upload commit 022984f323d3da44f70b3bf79c684cfd8dda3f61-dirty
nick
parents:
diff
changeset
|
104 for seq in seqs: |
af383638de66
planemo upload commit 022984f323d3da44f70b3bf79c684cfd8dda3f61-dirty
nick
parents:
diff
changeset
|
105 i+=1 |
af383638de66
planemo upload commit 022984f323d3da44f70b3bf79c684cfd8dda3f61-dirty
nick
parents:
diff
changeset
|
106 header = '>{}\n'.format(i) |
af383638de66
planemo upload commit 022984f323d3da44f70b3bf79c684cfd8dda3f61-dirty
nick
parents:
diff
changeset
|
107 family_file.write(header) |
af383638de66
planemo upload commit 022984f323d3da44f70b3bf79c684cfd8dda3f61-dirty
nick
parents:
diff
changeset
|
108 family_file.write(seq+'\n') |
af383638de66
planemo upload commit 022984f323d3da44f70b3bf79c684cfd8dda3f61-dirty
nick
parents:
diff
changeset
|
109 with open(os.devnull, 'w') as devnull: |
af383638de66
planemo upload commit 022984f323d3da44f70b3bf79c684cfd8dda3f61-dirty
nick
parents:
diff
changeset
|
110 try: |
af383638de66
planemo upload commit 022984f323d3da44f70b3bf79c684cfd8dda3f61-dirty
nick
parents:
diff
changeset
|
111 command = ['mafft', '--nuc', '--quiet', family_file.name] |
af383638de66
planemo upload commit 022984f323d3da44f70b3bf79c684cfd8dda3f61-dirty
nick
parents:
diff
changeset
|
112 output = subprocess.check_output(command, stderr=devnull) |
af383638de66
planemo upload commit 022984f323d3da44f70b3bf79c684cfd8dda3f61-dirty
nick
parents:
diff
changeset
|
113 except (OSError, subprocess.CalledProcessError): |
af383638de66
planemo upload commit 022984f323d3da44f70b3bf79c684cfd8dda3f61-dirty
nick
parents:
diff
changeset
|
114 return None |
af383638de66
planemo upload commit 022984f323d3da44f70b3bf79c684cfd8dda3f61-dirty
nick
parents:
diff
changeset
|
115 os.remove(family_file.name) |
af383638de66
planemo upload commit 022984f323d3da44f70b3bf79c684cfd8dda3f61-dirty
nick
parents:
diff
changeset
|
116 return read_fasta(output) |
af383638de66
planemo upload commit 022984f323d3da44f70b3bf79c684cfd8dda3f61-dirty
nick
parents:
diff
changeset
|
117 |
af383638de66
planemo upload commit 022984f323d3da44f70b3bf79c684cfd8dda3f61-dirty
nick
parents:
diff
changeset
|
118 |
af383638de66
planemo upload commit 022984f323d3da44f70b3bf79c684cfd8dda3f61-dirty
nick
parents:
diff
changeset
|
119 def read_fasta(fasta): |
af383638de66
planemo upload commit 022984f323d3da44f70b3bf79c684cfd8dda3f61-dirty
nick
parents:
diff
changeset
|
120 """Quick and dirty FASTA parser. Return only the list of sequences (no names). |
af383638de66
planemo upload commit 022984f323d3da44f70b3bf79c684cfd8dda3f61-dirty
nick
parents:
diff
changeset
|
121 Warning: Reads the entire contents of the file into memory at once.""" |
af383638de66
planemo upload commit 022984f323d3da44f70b3bf79c684cfd8dda3f61-dirty
nick
parents:
diff
changeset
|
122 sequences = [] |
af383638de66
planemo upload commit 022984f323d3da44f70b3bf79c684cfd8dda3f61-dirty
nick
parents:
diff
changeset
|
123 sequence = '' |
af383638de66
planemo upload commit 022984f323d3da44f70b3bf79c684cfd8dda3f61-dirty
nick
parents:
diff
changeset
|
124 for line in fasta.splitlines(): |
af383638de66
planemo upload commit 022984f323d3da44f70b3bf79c684cfd8dda3f61-dirty
nick
parents:
diff
changeset
|
125 if line.startswith('>'): |
af383638de66
planemo upload commit 022984f323d3da44f70b3bf79c684cfd8dda3f61-dirty
nick
parents:
diff
changeset
|
126 if sequence: |
af383638de66
planemo upload commit 022984f323d3da44f70b3bf79c684cfd8dda3f61-dirty
nick
parents:
diff
changeset
|
127 sequences.append(sequence) |
af383638de66
planemo upload commit 022984f323d3da44f70b3bf79c684cfd8dda3f61-dirty
nick
parents:
diff
changeset
|
128 sequence = '' |
af383638de66
planemo upload commit 022984f323d3da44f70b3bf79c684cfd8dda3f61-dirty
nick
parents:
diff
changeset
|
129 continue |
af383638de66
planemo upload commit 022984f323d3da44f70b3bf79c684cfd8dda3f61-dirty
nick
parents:
diff
changeset
|
130 sequence += line.strip() |
af383638de66
planemo upload commit 022984f323d3da44f70b3bf79c684cfd8dda3f61-dirty
nick
parents:
diff
changeset
|
131 if sequence: |
af383638de66
planemo upload commit 022984f323d3da44f70b3bf79c684cfd8dda3f61-dirty
nick
parents:
diff
changeset
|
132 sequences.append(sequence) |
af383638de66
planemo upload commit 022984f323d3da44f70b3bf79c684cfd8dda3f61-dirty
nick
parents:
diff
changeset
|
133 return sequences |
af383638de66
planemo upload commit 022984f323d3da44f70b3bf79c684cfd8dda3f61-dirty
nick
parents:
diff
changeset
|
134 |
af383638de66
planemo upload commit 022984f323d3da44f70b3bf79c684cfd8dda3f61-dirty
nick
parents:
diff
changeset
|
135 |
af383638de66
planemo upload commit 022984f323d3da44f70b3bf79c684cfd8dda3f61-dirty
nick
parents:
diff
changeset
|
136 def format_alignment(cons, seqs, quals=(), qual_thres=32, id_char='.'): |
af383638de66
planemo upload commit 022984f323d3da44f70b3bf79c684cfd8dda3f61-dirty
nick
parents:
diff
changeset
|
137 output = [cons.upper()] |
af383638de66
planemo upload commit 022984f323d3da44f70b3bf79c684cfd8dda3f61-dirty
nick
parents:
diff
changeset
|
138 for i, seq in enumerate(seqs): |
af383638de66
planemo upload commit 022984f323d3da44f70b3bf79c684cfd8dda3f61-dirty
nick
parents:
diff
changeset
|
139 outseq = '' |
af383638de66
planemo upload commit 022984f323d3da44f70b3bf79c684cfd8dda3f61-dirty
nick
parents:
diff
changeset
|
140 for j, seq_base in enumerate(seq.upper()): |
af383638de66
planemo upload commit 022984f323d3da44f70b3bf79c684cfd8dda3f61-dirty
nick
parents:
diff
changeset
|
141 if quals and seq_base != '-' and ord(quals[i][j]) < qual_thres: |
af383638de66
planemo upload commit 022984f323d3da44f70b3bf79c684cfd8dda3f61-dirty
nick
parents:
diff
changeset
|
142 outseq += ' ' |
af383638de66
planemo upload commit 022984f323d3da44f70b3bf79c684cfd8dda3f61-dirty
nick
parents:
diff
changeset
|
143 elif cons[j] == seq_base: |
af383638de66
planemo upload commit 022984f323d3da44f70b3bf79c684cfd8dda3f61-dirty
nick
parents:
diff
changeset
|
144 outseq += id_char |
af383638de66
planemo upload commit 022984f323d3da44f70b3bf79c684cfd8dda3f61-dirty
nick
parents:
diff
changeset
|
145 else: |
af383638de66
planemo upload commit 022984f323d3da44f70b3bf79c684cfd8dda3f61-dirty
nick
parents:
diff
changeset
|
146 outseq += seq_base |
af383638de66
planemo upload commit 022984f323d3da44f70b3bf79c684cfd8dda3f61-dirty
nick
parents:
diff
changeset
|
147 output.append(outseq) |
af383638de66
planemo upload commit 022984f323d3da44f70b3bf79c684cfd8dda3f61-dirty
nick
parents:
diff
changeset
|
148 return output |
af383638de66
planemo upload commit 022984f323d3da44f70b3bf79c684cfd8dda3f61-dirty
nick
parents:
diff
changeset
|
149 |
af383638de66
planemo upload commit 022984f323d3da44f70b3bf79c684cfd8dda3f61-dirty
nick
parents:
diff
changeset
|
150 |
af383638de66
planemo upload commit 022984f323d3da44f70b3bf79c684cfd8dda3f61-dirty
nick
parents:
diff
changeset
|
151 def fail(message): |
af383638de66
planemo upload commit 022984f323d3da44f70b3bf79c684cfd8dda3f61-dirty
nick
parents:
diff
changeset
|
152 sys.stderr.write(message+"\n") |
af383638de66
planemo upload commit 022984f323d3da44f70b3bf79c684cfd8dda3f61-dirty
nick
parents:
diff
changeset
|
153 sys.exit(1) |
af383638de66
planemo upload commit 022984f323d3da44f70b3bf79c684cfd8dda3f61-dirty
nick
parents:
diff
changeset
|
154 |
af383638de66
planemo upload commit 022984f323d3da44f70b3bf79c684cfd8dda3f61-dirty
nick
parents:
diff
changeset
|
155 if __name__ == '__main__': |
af383638de66
planemo upload commit 022984f323d3da44f70b3bf79c684cfd8dda3f61-dirty
nick
parents:
diff
changeset
|
156 sys.exit(main(sys.argv)) |