forked from kislyuk/bwa
-
Notifications
You must be signed in to change notification settings - Fork 0
/
bwa.py
410 lines (344 loc) · 18.9 KB
/
bwa.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
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
# This file is a part of bwa (DNAnexus platform app).
# Copyright (C) 2013 DNAnexus, Inc.
#
# This program is free software: you can redistribute it and/or modify it under
# the terms of the GNU General Public License as published by the Free Software
# Foundation; either version 2 of the License, or (at your option) any later
# version.
#
# This program is distributed in the hope that it will be useful, but WITHOUT
# ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or FITNESS
# FOR A PARTICULAR PURPOSE. See the GNU General Public License for more
# details.
#
# You should have received a copy of the GNU General Public License along with
# this program. If not, see <http://www.gnu.org/licenses/>.
import dxpy
import subprocess, logging, os, time, re
from multiprocessing import Pool, cpu_count
# Pypy specific workaround
os.environ['PYTHONPATH'] = os.environ.get('PYTHONPATH') + ":/usr/share/pyshared"
def run_shell(command):
logging.debug("Running "+command)
subprocess.check_call(command, shell=True)
def make_indexed_reference(job_inputs):
logging.info("Indexing reference genome")
run_shell("dx-contigset-to-fasta %s reference.fasta" % job_inputs['reference']['$dnanexus_link'])
ref_details = dxpy.DXRecord(job_inputs['reference']['$dnanexus_link']).get_details()
ref_name = dxpy.DXRecord(job_inputs['reference']['$dnanexus_link']).describe()['name']
# TODO: test if the genomes near the boundary work OK
if sum(ref_details['contigs']['sizes']) < 2*1024*1024*1024:
subprocess.check_call("bwa index -a is reference.fasta", shell=True)
else:
subprocess.check_call("bwa index -a bwtsw reference.fasta", shell=True)
subprocess.check_call("XZ_OPT=-0 tar -cJf reference.tar.xz reference.fasta*", shell=True)
indexed_ref_dxfile = dxpy.upload_local_file("reference.tar.xz", hidden=True, wait_on_close=True)
indexed_ref_record = dxpy.new_dxrecord(name=ref_name + " (indexed for BWA)",
types=["BwaLetterContigSetV3"],
details={'index_archive': dxpy.dxlink(indexed_ref_dxfile.get_id()),
'original_contigset': job_inputs['reference']})
indexed_ref_record.close()
# TODO: dxpy project workspace convenience functions
# FIXME
# if "projectWorkspace" in job:
# indexed_ref_record.clone(job["projectWorkspace"])
return indexed_ref_record
@dxpy.entry_point('main')
def main(**job_inputs):
job_outputs = {}
reads_inputs = job_inputs['reads']
reads_ids = [r['$dnanexus_link'] for r in reads_inputs]
reads_descriptions = {r: dxpy.DXGTable(r).describe() for r in reads_ids}
reads_columns = {r: [col['name'] for col in desc['columns']] for r, desc in reads_descriptions.items()}
print reads_inputs
print reads_ids
print reads_descriptions
print reads_columns
all_reads_have_FlowReads_tag = all(['FlowReads' in desc['types'] for desc in reads_descriptions.values()])
all_reads_have_LetterReads_tag = all(['LetterReads' in desc['types'] for desc in reads_descriptions.values()])
reads_have_names = any(['name' in columns for columns in reads_columns.values()])
reads_are_paired = any(['sequence2' in columns for columns in reads_columns.values()])
reads_have_qualities = any(['quality' in columns for columns in reads_columns.values()])
if reads_have_qualities:
assert(all(['quality' in columns for columns in reads_columns.values()]))
if reads_are_paired:
all_paired = all(['sequence2' in columns for columns in reads_columns.values()])
if not all_paired:
raise dxpy.AppError("Reads to be mapped must be either all paired or all unpaired. App input contains both paired and unpaired reads.")
if job_inputs["algorithm"] == "bwasw":
assert(not reads_are_paired) # bwasw does not support paired inputs
assert(all_reads_have_FlowReads_tag or all_reads_have_LetterReads_tag)
reference_record_types = dxpy.describe(job_inputs['reference'])['types']
if "BwaLetterContigSetV3" in reference_record_types:
input_ref_is_indexed = True
elif "ContigSet" in reference_record_types:
input_ref_is_indexed = False
else:
raise dxpy.ProgramError("Unrecognized object passed as reference. It must be a ContigSet record or a BwaLetterContigSetV3 file")
if input_ref_is_indexed:
job_outputs['indexed_reference'] = job_inputs['reference']
else:
found_cached_idx = False
for result in dxpy.find_data_objects(classname='record',
typename='BwaLetterContigSetV3',
link=job_inputs['reference']['$dnanexus_link']):
job_outputs['indexed_reference'] = dxpy.dxlink(result['id'])
found_cached_idx = True
break
if not found_cached_idx:
job_outputs['indexed_reference'] = dxpy.dxlink(make_indexed_reference(job_inputs))
table_columns = [("sequence", "string")]
if reads_have_names:
table_columns.append(("name", "string"))
if reads_have_qualities:
table_columns.append(("quality", "string"))
table_columns.extend([("status", "string"),
("chr", "string"),
("lo", "int32"),
("hi", "int32"),
("negative_strand", "boolean"),
("error_probability", "uint8"),
("qc_fail", "boolean"),
("duplicate", "boolean"),
("cigar", "string"),
("template_id", "int64"),
("read_group", "int32")])
# optional sam fields: RG BC XC XT NM CM XN SM AM XM X0 X1 XG MD XA
if reads_are_paired:
table_columns.extend([("mate_id", "int32"), # TODO: int8
("status2", "string"),
("chr2", "string"),
("lo2", "int32"),
("hi2", "int32"),
("negative_strand2", "boolean"),
("proper_pair", "boolean")])
if all_reads_have_FlowReads_tag:
table_columns.extend([("flowgram", "string"),
("flow_indices", "string"),
("clip_qual_left", "int32"),
("clip_qual_right", "int32"),
("clip_adapter_left", "int32"),
("clip_adapter_right", "int32")])
table_columns.extend([("sam_field_BC", "string"),
("sam_field_XC", "int32"),
("sam_field_XT", "string"),
("sam_field_NM", "int32"),
("sam_field_CM", "int32"),
("sam_field_XN", "int32"),
("sam_field_SM", "int32"),
("sam_field_AM", "int32"),
("sam_field_XM", "int32"),
("sam_field_X0", "int32"),
("sam_field_X1", "int32"),
("sam_field_XG", "int32"),
("sam_field_MD", "string"),
("sam_field_XA", "string"),
("sam_optional_fields", "string")])
column_descriptors = [dxpy.DXGTable.make_column_desc(name, type) for name, type in table_columns]
gri_index = dxpy.DXGTable.genomic_range_index("chr", "lo", "hi")
t = dxpy.new_dxgtable(column_descriptors, indices=[gri_index])
if input_ref_is_indexed:
original_contigset = dxpy.get_details(job_inputs['reference'])['original_contigset']
else:
original_contigset = job_inputs['reference']
t.set_details({'original_contigset': original_contigset})
t.add_types(["LetterMappings", "Mappings", "gri"])
# name table
if 'output_name' in job_inputs:
t.rename(job_inputs['output_name'])
else:
first_reads_name = dxpy.DXGTable( job_inputs['reads'][0] ).describe()['name']
contig_set_name = dxpy.describe(job_inputs['reference'])['name']
# if we're working on an indexed_reference we're not guaranteed to have access to original_contigset
if input_ref_is_indexed:
contig_set_name = contig_set_name.split(' (index')[0]
t.rename(first_reads_name + " mapped to " + contig_set_name)
# declare how many paired or single reads are in each reads table
read_group_lengths = []
for i in range(len(reads_ids)):
current_length = reads_descriptions[reads_ids[i]]["length"]
if 'sequence2' in dxpy.DXGTable(reads_ids[i]).get_col_names():
num_pairs = current_length
num_singles = 0
else:
num_pairs = 0
num_singles = current_length
read_group_lengths.append( {"num_singles":num_singles, "num_pairs":num_pairs} )
details = t.get_details()
details['read_groups'] = read_group_lengths
t.set_details(details)
row_offsets = []; row_cursor = 0
for i in range(len(reads_ids)):
row_offsets.append(row_cursor)
row_cursor += reads_descriptions[reads_ids[i]]["length"]
chunk_size = job_inputs["chunk_size"]
map_job_inputs = job_inputs.copy()
map_job_inputs["row_offsets"] = row_offsets
map_job_inputs["num_rows"] = chunk_size
map_job_inputs["table_id"] = t.get_id()
map_job_inputs["indexed_reference"] = job_outputs['indexed_reference']
postprocess_job_inputs = job_inputs.copy()
postprocess_job_inputs["table_id"] = t.get_id()
for start_row in xrange(0, row_cursor, chunk_size):
map_job_inputs["start_row"] = start_row
map_job = dxpy.new_dxjob(map_job_inputs, "map")
print "Launched map job with", map_job_inputs
postprocess_job_inputs["chunk%dresult" % start_row] = {'job': map_job.get_id(), 'field': 'ok'}
postprocess_job_inputs["chunk%ddebug" % start_row] = {'job': map_job.get_id(), 'field': 'debug'}
postprocess_job = dxpy.new_dxjob(postprocess_job_inputs, "postprocess")
job_outputs['mappings'] = {'job': postprocess_job.get_id(), 'field': 'mappings'}
print "MAIN OUTPUT:", job_outputs
return job_outputs
def write_reads_to_fastq(reads_id, filename, seq_col='sequence', qual_col='quality', start_row=0, end_row=None):
row_id = start_row
with open(filename, "w") as fh:
for row in dxpy.open_dxgtable(reads_id).iterate_rows(columns=[seq_col, qual_col], start=start_row, end=end_row):
for line in '@%d' % row_id, row[0], "+", row[1]:
print >>fh, line
row_id += 1
def write_reads_to_fasta(reads_id, filename, seq_col='sequence', start_row=0, end_row=None):
row_id = start_row
with open(filename, "w") as fh:
for row in dxpy.open_dxgtable(reads_id).iterate_rows(columns=[seq_col], start=start_row, end=end_row):
for line in '>%d' % row_id, row[0]:
print >>fh, line
row_id += 1
def run_alignment(algorithm, reads_file1, reads_file2=None, aln_opts='', sampe_opts='', sw_opts='', samse_opts=''):
commands = []
if algorithm == "bwasw":
if reads_file2 is None:
commands.append("bwa bwasw reference.fasta {r1} {sw_opts} > {r1}.sam")
else: # Paired read data
commands.append("bwa bwasw reference.fasta {r1} {r2} {sw_opts} > {r1}.sam")
else: # algorithm is "aln"
commands.append("bwa aln reference.fasta {r1} {aln_opts} > {r1}.sai")
if reads_file2 is not None:
commands.append("bwa aln reference.fasta {r2} {aln_opts} > {r2}.sai")
commands.append("bwa sampe reference.fasta {r1}.sai {r2}.sai {r1} {r2} {sampe_opts} > {r1}.sam")
else:
commands.append("bwa samse reference.fasta {r1}.sai {r1} {samse_opts} > {r1}.sam")
for command in commands:
run_shell(command.format(r1=reads_file1, r2=reads_file2, aln_opts=aln_opts, sampe_opts=sampe_opts, sw_opts=sw_opts, samse_opts=samse_opts))
def parse_bwa_cmd_opts(input):
aln_opts, sampe_opts, sw_opts, samse_opts = '', '', '', ''
for opt in 'n', 'o', 'e', 'i', 'd', 'l', 'k', 'm', 'M', 'O', 'E', 'R', 'q':
if 'aln_'+opt in input:
aln_opts += " -"+opt+" "+str(input['aln_'+opt])
if input['aln_N']:
aln_opts += ' -N'
for opt in 'a', 'o', 'n', 'N', 'c':
if 'sampe_'+opt in input:
sampe_opts += " -"+opt+" "+str(input['sampe_'+opt])
if input['sampe_s']:
sampe_opts += ' -s'
if 'samse_n' in input:
samse_opts += ' -n ' + str(input['samse_n'])
for opt in 'a', 'b', 'q', 'r', 'w', 'm', 'T', 'c', 'z', 's', 'N':
if 'sw_'+opt in input:
sw_opts += " -"+opt+" "+str(input['sw_'+opt])
return aln_opts, sampe_opts, sw_opts, samse_opts
@dxpy.entry_point('map')
def map(**job_inputs):
print "Map:", job_inputs
job_outputs = {}
times = [('start', time.time())]
reads_inputs = job_inputs['reads']
reads_ids = [r['$dnanexus_link'] for r in reads_inputs]
reads_descriptions = {r: dxpy.DXGTable(r).describe() for r in reads_ids}
reads_columns = {r: [col['name'] for col in desc['columns']] for r, desc in reads_descriptions.items()}
reads_are_paired = any(['sequence2' in columns for columns in reads_columns.values()])
times.append(('preamble', time.time()))
dxpy.download_dxfile(dxpy.get_details(job_inputs["indexed_reference"])['index_archive'], "reference.tar.xz")
times.append(('download reference', time.time()))
# TODO: Async everything below
# subprocess.check_call("pixz -d reference.tar.xz && tar -xf reference.tar", shell=True)
subprocess.check_call("tar -xJf reference.tar.xz", shell=True)
if job_inputs["algorithm"] == "bwasw":
bwa_algorithm = "bwasw"
else:
# algorithm = aln or auto. TODO: check what auto should do
bwa_algorithm = "aln"
aln_opts, sampe_opts, sw_opts, samse_opts = parse_bwa_cmd_opts(job_inputs)
# Set the number of threads BWA parameter to the apparent number of CPUs.
aln_opts += " -t " + str(cpu_count())
sw_opts += " -t " + str(cpu_count())
row_offsets = job_inputs['row_offsets'] # starting row for each reads table if you added them all up
start_row = job_inputs['start_row'] # the position in this chunk relative to the row_offsets 'total'
num_rows = job_inputs['num_rows'] # size of chunk to do this time
subjobs = []
for i in range(len(reads_ids)):
reads_length = reads_descriptions[reads_ids[i]]["length"]
read_group = i
# see if the reads table is part of this chunk
# if start is inside this reads table, add it
# doing this in the form: (A_start < B_end) and (A_end > B_start)
# A is the reads tables
# B is the current chunk
# A_start = row_offsets[i]
# A_end = row_offsets[i] + reads_length
# B_start = start_row
# B_end = start_row + num_rows
if row_offsets[i] < (start_row+num_rows) and (row_offsets[i]+reads_length) > start_row:
rel_start = max(start_row - row_offsets[i], 0)
rel_end = min(reads_length, start_row - row_offsets[i] + num_rows) # Using half-open intervals: [start, end)
subjobs.append({'reads_id': reads_ids[i], 'start_row': rel_start, 'end_row': rel_end, 'read_group':read_group})
times.append(('parse parameters', time.time()))
print 'SUBJOBS:', subjobs
for subchunk_id in range(len(subjobs)):
subjob = subjobs[subchunk_id]
reads_id = subjob['reads_id']
# TODO: FlowReads trimming support
if 'quality' in reads_columns[reads_id]:
if reads_are_paired:
reads_file1 = "input"+str(subchunk_id)+"_1.fastq"
reads_file2 = "input"+str(subchunk_id)+"_2.fastq"
write_reads_to_fastq(reads_id, reads_file1, seq_col='sequence', qual_col='quality', start_row=subjob['start_row'], end_row=subjob['end_row'])
write_reads_to_fastq(reads_id, reads_file2, seq_col='sequence2', qual_col='quality2', start_row=subjob['start_row'], end_row=subjob['end_row'])
times.append(('fetch reads (subchunk %d)' % subchunk_id, time.time()))
run_alignment(bwa_algorithm, reads_file1, reads_file2, aln_opts=aln_opts, sampe_opts=sampe_opts, sw_opts=sw_opts, samse_opts=samse_opts)
times.append(('run alignment (subchunk %d)' % subchunk_id, time.time()))
else:
reads_file1 = "input"+str(subchunk_id)+".fastq"
write_reads_to_fastq(reads_id, reads_file1, start_row=subjob['start_row'], end_row=subjob['end_row'])
run_alignment(bwa_algorithm, reads_file1, aln_opts=aln_opts, sampe_opts=sampe_opts, sw_opts=sw_opts, samse_opts=samse_opts)
else: # No qualities, use plain fasta
if reads_are_paired:
reads_file1 = "input"+str(subchunk_id)+"_1.fasta"
reads_file2 = "input"+str(subchunk_id)+"_2.fasta"
write_reads_to_fasta(reads_id, reads_file1, seq_col='sequence', start_row=subjob['start_row'], end_row=subjob['end_row'])
write_reads_to_fasta(reads_id, reads_file2, seq_col='sequence2', start_row=subjob['start_row'], end_row=subjob['end_row'])
run_alignment(bwa_algorithm, reads_file1, reads_file2, aln_opts=aln_opts, sampe_opts=sampe_opts, sw_opts=sw_opts, samse_opts=samse_opts)
else:
reads_file1 = "input"+str(subchunk_id)+".fasta"
write_reads_to_fasta(reads_id, reads_file1, start_row=subjob['start_row'], end_row=subjob['end_row'])
run_alignment(bwa_algorithm, reads_file1, aln_opts=aln_opts, sampe_opts=sampe_opts, sw_opts=sw_opts, samse_opts=samse_opts)
times.append(('run alignment (subchunk %d)' % subchunk_id, time.time()))
cmd = "dx_storeSamAsMappingsTable_bwa"
cmd += " --alignments '%s.sam'" % reads_file1
cmd += " --table_id '%s'" % job_inputs["table_id"]
cmd += " --reads_id '%s'" % reads_id
cmd += " --start_row %d" % subjob['start_row']
cmd += " --read_group %d" % subjob['read_group']
if job_inputs.get('discard_unmapped_rows'):
cmd += " --discard_unmapped_rows"
run_shell(cmd)
times.append(('run table upload (subchunk %d)' % subchunk_id, time.time()))
job_outputs["ok"] = True
timing_report = {}
for i in range(len(times)-1):
timing_report[times[i+1][0]] = times[i+1][1] - times[i][1]
job_outputs["debug"] = {'times': timing_report}
return job_outputs
@dxpy.entry_point('postprocess')
def postprocess(**job_inputs):
print "Postprocess:", job_inputs
job_outputs = {}
time_report = {k: v for k, v in job_inputs.iteritems() if re.match("chunk\d+debug", k)}
t = dxpy.DXGTable(job_inputs["table_id"])
d = t.get_details()
d['time_report'] = time_report
t.set_details(d)
t.close()
job_outputs['mappings'] = dxpy.dxlink(t)
return job_outputs
dxpy.run()