-
Notifications
You must be signed in to change notification settings - Fork 294
/
normalize-by-median.py
executable file
·366 lines (298 loc) · 13.9 KB
/
normalize-by-median.py
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
#! /usr/bin/env python2
#
# This file is part of khmer, http://github.com/ged-lab/khmer/, and is
# Copyright (C) Michigan State University, 2009-2015. It is licensed under
# the three-clause BSD license; see LICENSE.
# Contact: khmer-project@idyll.org
#
# pylint: disable=invalid-name,missing-docstring
"""
Eliminate surplus reads.
Eliminate reads with median k-mer abundance higher than
DESIRED_COVERAGE. Output sequences will be placed in 'infile.keep', with the
option to output to STDOUT.
% python scripts/normalize-by-median.py [ -C <cutoff> ] <data1> <data2> ...
Use '-h' for parameter help.
"""
import sys
import screed
import os
import khmer
import textwrap
from itertools import izip
from khmer.khmer_args import (build_counting_args, add_loadhash_args,
report_on_config, info)
import argparse
from khmer.kfile import (check_space, check_space_for_hashtable,
check_valid_file_exists)
from khmer.utils import write_record, check_is_pair
DEFAULT_DESIRED_COVERAGE = 10
# Iterate a collection in arbitrary batches
# from: http://stackoverflow.com/questions/4628290/pairs-from-single-list
def batchwise(coll, size):
iter_coll = iter(coll)
return izip(*[iter_coll] * size)
# Returns true if the pair of records are properly pairs
# pylint: disable=too-many-locals,too-many-branches
def normalize_by_median(input_filename, outfp, htable, paired, cutoff,
report_fp=None):
desired_coverage = cutoff
ksize = htable.ksize()
# In paired mode we read two records at a time
batch_size = 1
if paired:
batch_size = 2
index = -1
total = 0
discarded = 0
for index, batch in enumerate(batchwise(screed.open(
input_filename, parse_description=False), batch_size)):
if index > 0 and index % 100000 == 0:
print >>sys.stderr, '... kept {kept} of {total} or'\
' {perc:2}%'.format(kept=total - discarded, total=total,
perc=int(100. - discarded /
float(total) * 100.))
print >>sys.stderr, '... in file', input_filename
if report_fp:
print >> report_fp, total, total - discarded, \
1. - (discarded / float(total))
report_fp.flush()
total += batch_size
# If in paired mode, check that the reads are properly interleaved
if paired:
if not check_is_pair(batch[0], batch[1]):
raise IOError('Error: Improperly interleaved pairs \
{b0} {b1}'.format(b0=batch[0].name, b1=batch[1].name))
# Emit the batch of reads if any read passes the filter
# and all reads are longer than K
passed_filter = False
passed_length = True
for record in batch:
if len(record.sequence) < ksize:
passed_length = False
continue
seq = record.sequence.replace('N', 'A')
med, _, _ = htable.get_median_count(seq)
if med < desired_coverage:
htable.consume(seq)
passed_filter = True
# Emit records if any passed
if passed_length and passed_filter:
for record in batch:
write_record(record, outfp)
else:
discarded += batch_size
if report_fp:
print >> report_fp, total, total - discarded, \
1. - (discarded / float(total))
report_fp.flush()
return total, discarded
def handle_error(error, output_name, input_name, fail_save, htable):
print >> sys.stderr, '** ERROR:', error
print >> sys.stderr, '** Failed on {name}: '.format(name=input_name)
if fail_save:
tablename = os.path.basename(input_name) + '.ct.failed'
print >> sys.stderr, \
'** ...dumping k-mer counting table to {tn}'.format(tn=tablename)
htable.save(tablename)
try:
os.remove(output_name)
except: # pylint: disable=bare-except
print >> sys.stderr, '** ERROR: problem removing corrupt filtered file'
def normalize_by_median_and_check(input_filename, htable, single_output_file,
fail_save, paired, cutoff, force,
corrupt_files, report_fp=None):
total = 0
discarded = 0
total_acc = None
discarded_acc = None
if single_output_file:
if single_output_file is sys.stdout:
output_name = '/dev/stdout'
else:
output_name = single_output_file.name
outfp = single_output_file
else:
output_name = os.path.basename(input_filename) + '.keep'
outfp = open(output_name, 'w')
try:
total_acc, discarded_acc = normalize_by_median(
input_filename, outfp, htable, paired, cutoff, report_fp=None)
except IOError as err:
handle_error(err, output_name, input_filename, fail_save,
htable)
if not force:
print >> sys.stderr, '** Exiting!'
sys.exit(1)
else:
print >> sys.stderr, '*** Skipping error file, moving on...'
corrupt_files.append(input_filename)
else:
if total_acc == 0 and discarded_acc == 0:
print >> sys.stderr, 'SKIPPED empty file', input_filename
else:
total += total_acc
discarded += discarded_acc
print >> sys.stderr, \
'DONE with {inp}; kept {kept} of {total} or {perc:2}%'\
.format(inp=input_filename, kept=total - discarded,
total=total, perc=int(100. - discarded /
float(total) * 100.))
print >> sys.stderr, 'output in', output_name
return total_acc, discarded_acc, corrupt_files
def get_parser():
epilog = ("""
Discard sequences based on whether or not their median k-mer abundance lies
above a specified cutoff. Kept sequences will be placed in <fileN>.keep.
Paired end reads will be considered together if :option:`-p` is set. If
either read will be kept, then both will be kept. This should result in
keeping (or discarding) each sequencing fragment. This helps with retention
of repeats, especially. With :option: `-u`/:option:`--unpaired-reads`,
unpaired reads from the specified file will be read after the paired data
is read.
With :option:`-s`/:option:`--savetable`, the k-mer counting table
will be saved to the specified file after all sequences have been
processed. With :option:`-d`, the k-mer counting table will be
saved every d files for multifile runs; if :option:`-s` is set,
the specified name will be used, and if not, the name `backup.ct`
will be used. :option:`-l`/:option:`--loadtable` will load the
specified k-mer counting table before processing the specified
files. Note that these tables are are in the same format as those
produced by :program:`load-into-counting.py` and consumed by
:program:`abundance-dist.py`.
:option:`-f`/:option:`--fault-tolerant` will force the program to continue
upon encountering a formatting error in a sequence file; the k-mer counting
table up to that point will be dumped, and processing will continue on the
next file.
To append reads to an output file (rather than overwriting it), send output
to STDOUT with `--out -` and use UNIX file redirection syntax (`>>`) to
append to the file.
Example::
normalize-by-median.py -k 17 tests/test-data/test-abund-read-2.fa
Example::
""" " normalize-by-median.py -p -k 17 tests/test-data/test-abund-read-paired.fa" # noqa
"""
Example::
""" " normalize-by-median.py -p -k 17 -o - tests/test-data/paired.fq >> appended-output.fq" # noqa
"""
Example::
""" " normalize-by-median.py -k 17 -f tests/test-data/test-error-reads.fq tests/test-data/test-fastq-reads.fq" # noqa
"""
Example::
""" " normalize-by-median.py -k 17 -d 2 -s test.ct tests/test-data/test-abund-read-2.fa tests/test-data/test-fastq-reads") # noqa
parser = build_counting_args(
descr="Do digital normalization (remove mostly redundant sequences)",
epilog=textwrap.dedent(epilog))
parser.add_argument('-C', '--cutoff', type=int,
default=DEFAULT_DESIRED_COVERAGE)
parser.add_argument('-p', '--paired', action='store_true')
parser.add_argument('-u', '--unpaired-reads',
metavar="unpaired_reads_filename", help='with paired data only,\
include an unpaired file')
parser.add_argument('-s', '--savetable', metavar="filename", default='',
help='save the k-mer counting table to disk after all'
'reads are loaded.')
parser.add_argument('-R', '--report',
metavar='filename', type=argparse.FileType('w'))
parser.add_argument('-f', '--fault-tolerant', dest='force',
help='continue on next file if read errors are \
encountered', action='store_true')
parser.add_argument('--save-on-failure', dest='fail_save',
action='store_false', default=True,
help='Save k-mer counting table when an error occurs')
parser.add_argument('-d', '--dump-frequency', dest='dump_frequency',
type=int, help='dump k-mer counting table every d '
'files', default=-1)
parser.add_argument('-o', '--out', metavar="filename",
dest='single_output_file',
type=argparse.FileType('w'),
default=None, help='only output a single file with '
'the specified filename; use a single dash "-" to '
'specify that output should go to STDOUT (the '
'terminal)')
parser.add_argument('input_filenames', metavar='input_sequence_filename',
help='Input FAST[AQ] sequence filename.', nargs='+')
parser.add_argument('--report-total-kmers', '-t', action='store_true',
help="Prints the total number of k-mers"
" post-normalization to stderr")
parser.add_argument('--force', default=False, action='store_true',
help='Overwrite output file if it exists')
add_loadhash_args(parser)
return parser
def main(): # pylint: disable=too-many-branches,too-many-statements
info('normalize-by-median.py', ['diginorm'])
args = get_parser().parse_args()
report_on_config(args)
report_fp = args.report
# check for similar filenames
filenames = []
for pathfilename in args.input_filenames:
filename = pathfilename.split('/')[-1]
if (filename in filenames):
print >>sys.stderr, "WARNING: At least two input files are named \
%s . (The script normalize-by-median.py can not handle this, only one .keep \
file for one of the input files will be generated.)" % filename
else:
filenames.append(filename)
# check for others
check_valid_file_exists(args.input_filenames)
check_space(args.input_filenames, args.force)
if args.savetable:
check_space_for_hashtable(
args.n_tables * args.min_tablesize, args.force)
# list to save error files along with throwing exceptions
corrupt_files = []
if args.loadtable:
print 'loading k-mer counting table from', args.loadtable
htable = khmer.load_counting_hash(args.loadtable)
else:
print >> sys.stderr, 'making k-mer counting table'
htable = khmer.new_counting_hash(args.ksize, args.min_tablesize,
args.n_tables)
input_filename = None
for index, input_filename in enumerate(args.input_filenames):
total_acc, discarded_acc, corrupt_files = \
normalize_by_median_and_check(
input_filename, htable, args.single_output_file,
args.fail_save, args.paired, args.cutoff, args.force,
corrupt_files, report_fp)
if (args.dump_frequency > 0 and
index > 0 and index % args.dump_frequency == 0):
print 'Backup: Saving k-mer counting file through', input_filename
if args.savetable:
hashname = args.savetable
print '...saving to', hashname
else:
hashname = 'backup.ct'
print 'Nothing given for savetable, saving to', hashname
htable.save(hashname)
if args.paired and args.unpaired_reads:
args.paired = False
output_name = args.unpaired_reads
if not args.single_output_file:
output_name = os.path.basename(args.unpaired_reads) + '.keep'
outfp = open(output_name, 'w')
total_acc, discarded_acc, corrupt_files = \
normalize_by_median_and_check(
args.unpaired_reads, htable, args.single_output_file,
args.fail_save, args.paired, args.cutoff, args.force,
corrupt_files, report_fp)
if args.report_total_kmers:
print >> sys.stderr, 'Total number of unique k-mers: {0}'.format(
htable.n_unique_kmers())
if args.savetable:
print 'Saving k-mer counting table through', input_filename
print '...saving to', args.savetable
htable.save(args.savetable)
fp_rate = \
khmer.calc_expected_collisions(htable, args.force, max_false_pos=.8)
# for max_false_pos see Zhang et al., http://arxiv.org/abs/1309.2975
print >> sys.stderr, \
'fp rate estimated to be {fpr:1.3f}'.format(fpr=fp_rate)
if args.force and len(corrupt_files) > 0:
print >> sys.stderr, "** WARNING: Finished with errors!"
print >> sys.stderr, "** IOErrors occurred in the following files:"
print >> sys.stderr, "\t", " ".join(corrupt_files)
if __name__ == '__main__':
main()
# vim: set ft=python ts=4 sts=4 sw=4 et tw=79: