def test_get_num_threads():
'tests get_num_threads'
threads = 3
assert get_num_threads(threads) == threads
threads = False
assert get_num_threads(threads) == 1
threads = True
assert get_num_threads(threads) == os.sysconf('SC_NPROCESSORS_ONLN')
threads = True
limit_by_memory = 1024
assert 1 <= get_num_threads(threads, limit_by_memory) <= os.sysconf('SC_NPROCESSORS_ONLN')
def realign_bam(bam_fpath, reference_fpath, out_bam_fpath, java_conf=None,
threads=False, tmp_dir=None):
'It realigns the bam using GATK Local realignment around indels'
#reference sam index
create_sam_reference_index(reference_fpath)
#reference picard dict
create_picard_dict(reference_fpath, java_conf=java_conf)
#bam index
create_bam_index(bam_fpath)
#the intervals to realign
gatk_path = guess_jar_dir('GenomeAnalysisTK.jar', java_conf)
gatk_jar = os.path.join(gatk_path, 'GenomeAnalysisTK.jar')
intervals_fhand = tempfile.NamedTemporaryFile(suffix='.intervals')
cmd = java_cmd(java_conf=java_conf)
cmd.extend(['-jar', gatk_jar, '-T', 'RealignerTargetCreator',
'-I', bam_fpath, '-R', reference_fpath, '-o', intervals_fhand.name])
#according to GATK this is experimental, so it might be a good idea to
#do it in just one thread. In version 1.0.4498. This options is removed
# so parallel = False
parallel = False
if parallel and threads and threads > 1:
cmd.extend(['-nt', str(get_num_threads(threads))])
call(cmd, raise_on_error=True, add_ext_dir=False)
#the realignment itself
unsorted_bam = NamedTemporaryFile(suffix='.bam')
cmd = java_cmd(java_conf=java_conf)
cmd.extend(['-Djava.io.tmpdir=%s' % tempfile.gettempdir(),
'-jar', gatk_jar, '-I', bam_fpath, '-R', reference_fpath,
'-T', 'IndelRealigner', '-targetIntervals', intervals_fhand.name,
'-o', unsorted_bam.name])
if parallel and threads and threads > 1:
cmd.extend(['-nt', str(get_num_threads(threads))])
call(cmd, raise_on_error=True, add_ext_dir=False)
# now we have to realign the bam
sort_bam_sam(unsorted_bam.name, out_bam_fpath, java_conf=java_conf,
tmp_dir=tmp_dir)
def map_reads_with_bwa(reference_fpath, reads_fpath, bam_fpath, parameters):
'It maps the reads to the reference using bwa and returns a bam file'
colorspace = parameters['colorspace']
reads_length = parameters['reads_length']
threads = parameters['threads']
java_conf = parameters['java_conf']
tmp_dir = parameters['tmp_dir'] if 'tmp_dir' in parameters else None
threads = get_num_threads(threads)
#the reference should have an index
bwt_fpath = reference_fpath + '.bwt'
if not os.path.exists(bwt_fpath):
create_bwa_reference(reference_fpath, color=colorspace)
output_ali = 'output.ali'
bam_file_bam = 'bam_file.bam'
output_sai = 'output.sai'
if reads_length == 'short':
cmd = ['bwa', 'aln', reference_fpath, reads_fpath,
'-t', str(threads)]
if colorspace:
cmd.append('-c')
sai_fhand = NamedTemporaryFile(dir=tmp_dir, suffix=output_sai, mode='wb')
call(cmd, stdout=sai_fhand, raise_on_error=True)
cmd = ['bwa', 'samse', reference_fpath, sai_fhand.name, reads_fpath]
ali_fhand = NamedTemporaryFile(dir=tmp_dir, suffix=output_ali, mode='w')
call(cmd, stdout=ali_fhand, raise_on_error=True)
elif reads_length == 'long':
cmd = ['bwa', 'dbwtsw', reference_fpath, reads_fpath,
'-t', str(threads)]
ali_fhand = NamedTemporaryFile(dir=tmp_dir, suffix=output_ali)
call(cmd, stdout=ali_fhand, raise_on_error=True)
else:
raise ValueError('Reads length: short or long')
if 'unmapped_fhand' in parameters and parameters['unmapped_fhand'] is not None:
out_ali_fhand = NamedTemporaryFile(dir=tmp_dir, suffix=output_ali)
get_out_unmapped(ali_fhand, parameters['unmapped_fhand'], out_ali_fhand)
ali_fhand = out_ali_fhand
# From sam to Bam
# unsorted_bam = os.path.join(temp_dir.name, bam_file_bam)
unsorted_bam = NamedTemporaryFile(dir=tmp_dir, suffix=bam_file_bam)
sam2bam(ali_fhand.name, unsorted_bam.name)
# sort bam file
sort_bam_sam(unsorted_bam.name, bam_fpath, sort_method='coordinate',
java_conf=java_conf, strict_validation=False, tmp_dir=tmp_dir)
def _get_num_threads(self):
'It calculates the number of threads to use'
threads = self._project_settings['General_settings']['threads']
return get_num_threads(threads)
def _get_num_threads(self):
"It calculates the number of threads to use"
threads = self._project_settings["General_settings"]["threads"]
return get_num_threads(threads)
def _get_num_threads(self):
'It calculates the number of threads for snv calling analisys'
threads = self._project_settings['General_settings']['threads']
return get_num_threads(threads, PYSAM_MEMORY_LIMIT)