Mercurial > repos > cstrittmatter > test_eurl_vtec_wgs_pt

annotate scripts/ReMatCh/utils/strip_alignment.py @ 3:0cbed1c0a762 draft default tip

Find changesets by keywords (author, files, the commit message), revision number or hash, or revset expression.
planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
author cstrittmatter
date Tue, 28 Jan 2020 10:42:31 -0500
parents 965517909457
children
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
rev   line source
3
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents: 0
diff changeset
1 #!/usr/bin/env python3
0
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
2
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
3 # -*- coding: utf-8 -*-
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
4
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
5 """
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
6 strip_alignment.py - Strip alignment positions containing gaps,
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
7 missing data and invariable positions
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
8 <https://github.com/B-UMMI/ReMatCh/>
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
9
3
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents: 0
diff changeset
10 Copyright (C) 2018 Miguel Machado <mpmachado@medicina.ulisboa.pt>
0
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
11
3
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents: 0
diff changeset
12 Last modified: October 15, 2018
0
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
13
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
14 This program is free software: you can redistribute it and/or modify
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
15 it under the terms of the GNU General Public License as published by
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
16 the Free Software Foundation, either version 3 of the License, or
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
17 (at your option) any later version.
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
18
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
19 This program is distributed in the hope that it will be useful,
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
20 but WITHOUT ANY WARRANTY; without even the implied warranty of
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
21 MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
22 GNU General Public License for more details.
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
23
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
24 You should have received a copy of the GNU General Public License
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
25 along with this program. If not, see <http://www.gnu.org/licenses/>.
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
26 """
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
27
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
28 from Bio import SeqIO
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
29 import os
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
30 import argparse
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
31 import sys
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
32
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
33
3
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents: 0
diff changeset
34 version = '0.2'
0
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
35
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
36
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
37 def get_sequences(infile):
3
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents: 0
diff changeset
38 print('Getting sequences')
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents: 0
diff changeset
39 sequences_seq_io = list(SeqIO.parse(infile, 'fasta'))
0
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
40
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
41 sequence_length = None
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
42 sequences_dict = {}
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
43 all_executed_printed = False
3
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents: 0
diff changeset
44 for x, sequence in enumerate(sequences_seq_io):
0
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
45 if sequence_length is None:
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
46 sequence_length = len(sequence.seq)
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
47 if sequence_length != len(sequence.seq):
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
48 sys.exit('Sequences with different length!')
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
49 sequences_dict[sequence.id] = list(sequence.seq)
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
50
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
51 if (x + 1) % 10 == 0:
3
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents: 0
diff changeset
52 print('\n' + str(round((float(x + 1) / len(sequences_seq_io)) * 100, 2)) + '% of sequences already processed (getting sequences)')
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents: 0
diff changeset
53 if x + 1 == len(sequences_seq_io):
0
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
54 all_executed_printed = True
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
55 if not all_executed_printed:
3
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents: 0
diff changeset
56 print('\n' + str(round((float(x + 1) / len(sequences_seq_io)) * 100, 2)) + '% of sequences already processed (getting sequences)')
0
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
57
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
58 return sequences_dict, sequence_length
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
59
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
60
3
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents: 0
diff changeset
61 def positions_type(sequences_dict, sequence_length, not_gaps, not_missing, not_invariable):
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents: 0
diff changeset
62 print('Determining positions type')
0
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
63 positions_2_keep = []
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
64 invariable = []
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
65 missing = []
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
66 gaps = []
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
67 gaps_missing = 0
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
68 all_executed_printed = False
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
69 for i in range(0, sequence_length):
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
70 data = []
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
71 for sample in sequences_dict:
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
72 data.append(sequences_dict[sample][i])
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
73 possibilities = set(data)
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
74 if len(possibilities) == 1:
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
75 invariable.append(i)
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
76 if len(possibilities.intersection(set(['N']))) > 0:
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
77 missing.append(i)
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
78 if len(possibilities.intersection(set(['-']))) > 0:
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
79 gaps.append(i)
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
80 if len(possibilities.intersection(set(['N', '-']))) > 0:
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
81 gaps_missing += 1
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
82 if len(possibilities) > 1 and len(possibilities.intersection(set(['N', '-']))) == 0:
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
83 positions_2_keep.append(i)
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
84
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
85 if (i + 1) % 10000 == 0:
3
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents: 0
diff changeset
86 print('\n' + str(round((float(i + 1) / sequence_length) * 100, 2)) + '% of positions already'
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents: 0
diff changeset
87 ' processed (determining positions'
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents: 0
diff changeset
88 ' type)')
0
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
89 if i + 1 == len(sequences_dict):
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
90 all_executed_printed = True
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
91 if not all_executed_printed:
3
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents: 0
diff changeset
92 print('\n' + str(round((float(i + 1) / sequence_length) * 100, 2)) + '% of positions already'
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents: 0
diff changeset
93 ' processed (determining positions'
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents: 0
diff changeset
94 ' type)')
0
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
95
3
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents: 0
diff changeset
96 print('Positions to keep (no matter): ' + str(len(positions_2_keep)))
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents: 0
diff changeset
97 print('Invariable sites: ' + str(len(invariable)))
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents: 0
diff changeset
98 print('Positions with missing data ("N"): ' + str(len(missing)))
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents: 0
diff changeset
99 print('Positions with GAPs ("-"): ' + str(len(gaps)))
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents: 0
diff changeset
100 print('Positions with GAPs or missing data: ' + str(gaps_missing))
0
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
101
3
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents: 0
diff changeset
102 if not_gaps:
0
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
103 positions_2_keep.extend(gaps)
3
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents: 0
diff changeset
104 if not_missing:
0
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
105 positions_2_keep.extend(missing)
3
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents: 0
diff changeset
106 if not_invariable:
0
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
107 positions_2_keep.extend(invariable)
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
108
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
109 positions_2_keep = sorted(set(positions_2_keep))
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
110
3
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents: 0
diff changeset
111 print('Positions to keep (final): ' + str(len(positions_2_keep)))
0
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
112
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
113 return positions_2_keep
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
114
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
115
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
116 def chunkstring(string, length):
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
117 return (string[0 + i:length + i] for i in range(0, len(string), length))
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
118
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
119
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
120 def write_fasta(sequences_dict, positions_2_keep, outfile):
3
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents: 0
diff changeset
121 print('Writing stripped sequences')
0
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
122 all_executed_printed = False
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
123 with open(outfile, 'wt') as writer:
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
124 for x, sample in enumerate(sequences_dict):
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
125 writer.write('>' + sample + '\n')
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
126 fasta_sequence_lines = chunkstring(''.join([sequences_dict[sample][i] for i in positions_2_keep]), 80)
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
127 for line in fasta_sequence_lines:
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
128 writer.write(line + '\n')
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
129
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
130 if (x + 1) % 100 == 0:
3
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents: 0
diff changeset
131 print('\n' + str(round((float(x + 1) / len(sequences_dict)) * 100, 2)) + '% of sequences already'
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents: 0
diff changeset
132 ' processed (writing stripped'
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents: 0
diff changeset
133 ' sequences)')
0
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
134 if x + 1 == len(sequences_dict):
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
135 all_executed_printed = True
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
136 if not all_executed_printed:
3
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents: 0
diff changeset
137 print('\n' + str(round((float(x + 1) / len(sequences_dict)) * 100, 2)) + '% of sequences already'
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents: 0
diff changeset
138 ' processed (writing stripped'
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents: 0
diff changeset
139 ' sequences)')
0
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
140
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
141
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
142 def strip_alignment(args):
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
143 outdir = os.path.dirname(os.path.abspath(args.outfile))
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
144 if not os.path.isdir(outdir):
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
145 os.makedirs(outdir)
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
146
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
147 outfile = os.path.abspath(args.outfile)
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
148
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
149 infile = os.path.abspath(args.infile.name)
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
150
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
151 sequences_dict, sequence_length = get_sequences(infile)
3
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents: 0
diff changeset
152 positions_2_keep = positions_type(sequences_dict, sequence_length, args.notGAPs, args.notMissing,
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents: 0
diff changeset
153 args.notInvariable)
0
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
154 write_fasta(sequences_dict, positions_2_keep, outfile)
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
155
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
156
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
157 def main():
3
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents: 0
diff changeset
158 parser = argparse.ArgumentParser(prog='strip_alignment.py',
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents: 0
diff changeset
159 description='Strip alignment positions containing gaps, missing data and'
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents: 0
diff changeset
160 ' invariable positions',
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents: 0
diff changeset
161 formatter_class=argparse.ArgumentDefaultsHelpFormatter)
0
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
162 parser.add_argument('--version', help='Version information', action='version', version=str('%(prog)s v' + version))
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
163
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
164 parser_required = parser.add_argument_group('Required options')
3
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents: 0
diff changeset
165 parser_required.add_argument('-i', '--infile', type=argparse.FileType('r'),
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents: 0
diff changeset
166 metavar='/path/to/aligned/input/file.fasta', help='Path to the aligned fasta file',
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents: 0
diff changeset
167 required=True)
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents: 0
diff changeset
168 parser_required.add_argument('-o', '--outfile', type=str, metavar='/path/to/stripped/output/file.fasta',
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents: 0
diff changeset
169 help='Stripped output fasta file', required=True, default='alignment_stripped.fasta')
0
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
170
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
171 parser_optional_general = parser.add_argument_group('General facultative options')
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
172 parser_optional_general.add_argument('--notGAPs', action='store_true', help='Not strip positions with GAPs')
3
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents: 0
diff changeset
173 parser_optional_general.add_argument('--notMissing', action='store_true',
0cbed1c0a762 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents: 0
diff changeset
174 help='Not strip positions with missing data')
0
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
175 parser_optional_general.add_argument('--notInvariable', action='store_true', help='Not strip invariable sites')
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
176
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
177 args = parser.parse_args()
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
178
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
179 strip_alignment(args)
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
180
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
181
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
182 if __name__ == "__main__":
965517909457 planemo upload commit 15239f1674081ab51ab8dd75a9a40cf1bfaa93e8
cstrittmatter
parents:
diff changeset
183 main()