def _piped_realign_gatk(data, region, cl, out_base_file, tmp_dir, prep_params):
"""Perform realignment with GATK, using input commandline.
GATK requires writing to disk and indexing before realignment.
"""
broad_runner = broad.runner_from_config(data["config"])
pa_bam = "%s-prealign%s" % os.path.splitext(out_base_file)
if not utils.file_exists(pa_bam):
with file_transaction(data, pa_bam) as tx_out_file:
cmd = "{cl} -o {tx_out_file}".format(**locals())
do.run(cmd, "GATK re-alignment {0}".format(region), data)
bam.index(pa_bam, data["config"])
recal_file = realign.gatk_realigner_targets(
broad_runner,
pa_bam,
data["sam_ref"],
data["config"],
region=region_to_gatk(region),
known_vrns=dd.get_variation_resources(data))
recal_cl = realign.gatk_indel_realignment_cl(
broad_runner,
pa_bam,
data["sam_ref"],
recal_file,
tmp_dir,
region=region_to_gatk(region),
known_vrns=dd.get_variation_resources(data))
return pa_bam, recal_cl
def _piped_realign_gatk(data, region, cl, out_base_file, tmp_dir):
"""Perform realignment with GATK, using input commandline.
GATK requires writing to disk and indexing before realignment.
"""
broad_runner = broad.runner_from_config(data["config"])
pa_bam = "%s-prealign%s" % os.path.splitext(out_base_file)
if not utils.file_exists(pa_bam):
with file_transaction(pa_bam) as tx_out_file:
subprocess.check_call("{cl} > {tx_out_file}".format(**locals()),
shell=True)
broad_runner.run_fn("picard_index", pa_bam)
dbsnp_vcf = shared.configured_ref_file("dbsnp", data["config"],
data["sam_ref"])
recal_file = realign.gatk_realigner_targets(broad_runner,
pa_bam,
data["sam_ref"],
dbsnp=dbsnp_vcf,
region=region_to_gatk(region))
recal_cl = realign.gatk_indel_realignment_cl(broad_runner,
pa_bam,
data["sam_ref"],
recal_file,
tmp_dir,
region=region_to_gatk(region))
return pa_bam, " ".join(recal_cl)
def _piped_realign_gatk(data, region, cl, out_base_file, tmp_dir, prep_params):
"""Perform realignment with GATK, using input commandline.
GATK requires writing to disk and indexing before realignment.
"""
broad_runner = broad.runner_from_config(data["config"])
pa_bam = "%s-prealign%s" % os.path.splitext(out_base_file)
if not utils.file_exists(pa_bam):
with file_transaction(pa_bam) as tx_out_file:
pipe = ">" if prep_params["dup"] else "-o"
cmd = "{cl} {pipe} {tx_out_file}".format(**locals())
do.run(cmd, "GATK pre-alignment {0}".format(region), data)
bam.index(pa_bam, data["config"])
dbsnp_vcf = data["genome_resources"]["variation"]["dbsnp"]
recal_file = realign.gatk_realigner_targets(broad_runner,
pa_bam,
data["sam_ref"],
dbsnp=dbsnp_vcf,
region=region_to_gatk(region))
recal_cl = realign.gatk_indel_realignment_cl(broad_runner,
pa_bam,
data["sam_ref"],
recal_file,
tmp_dir,
region=region_to_gatk(region))
return pa_bam, " ".join(recal_cl)
def _piped_realign_gatk(data, region, cl, out_base_file, tmp_dir, prep_params):
"""Perform realignment with GATK, using input commandline.
GATK requires writing to disk and indexing before realignment.
"""
broad_runner = broad.runner_from_config(data["config"])
pa_bam = "%s-prealign%s" % os.path.splitext(out_base_file)
if not utils.file_exists(pa_bam):
with file_transaction(data, pa_bam) as tx_out_file:
cmd = "{cl} -o {tx_out_file}".format(**locals())
do.run(cmd, "GATK pre-alignment {0}".format(region), data)
bam.index(pa_bam, data["config"])
recal_file = realign.gatk_realigner_targets(
broad_runner,
pa_bam,
data["sam_ref"],
data["config"],
region=region_to_gatk(region),
known_vrns=dd.get_variation_resources(data),
)
recal_cl = realign.gatk_indel_realignment_cl(
broad_runner,
pa_bam,
data["sam_ref"],
recal_file,
tmp_dir,
region=region_to_gatk(region),
known_vrns=dd.get_variation_resources(data),
)
return pa_bam, " ".join(recal_cl)
def _piped_realign_gatk(data, region, cl, out_base_file, tmp_dir):
"""Perform realignment with GATK, using input commandline.
GATK requires writing to disk and indexing before realignment.
"""
broad_runner = broad.runner_from_config(data["config"])
pa_bam = "%s-prealign%s" % os.path.splitext(out_base_file)
if not utils.file_exists(pa_bam):
with file_transaction(pa_bam) as tx_out_file:
subprocess.check_call("{cl} > {tx_out_file}".format(**locals()), shell=True)
broad_runner.run_fn("picard_index", pa_bam)
recal_file = realign.gatk_realigner_targets(broad_runner, pa_bam, data["sam_ref"],
dbsnp=shared.configured_ref_file("dbsnp", data["config"], data["sam_ref"]),
region=_region_to_gatk(region))
recal_cl = realign.gatk_indel_realignment_cl(broad_runner, pa_bam, data["sam_ref"],
recal_file, tmp_dir, region=_region_to_gatk(region))
return pa_bam, " ".join(recal_cl)
def _piped_realign_gatk(data, region, cl, out_base_file, tmp_dir, prep_params):
"""Perform realignment with GATK, using input commandline.
GATK requires writing to disk and indexing before realignment.
"""
broad_runner = broad.runner_from_config(data["config"])
pa_bam = "%s-prealign%s" % os.path.splitext(out_base_file)
if not utils.file_exists(pa_bam):
with file_transaction(pa_bam) as tx_out_file:
pipe = ">" if prep_params["dup"] else "-o"
cmd = "{cl} {pipe} {tx_out_file}".format(**locals())
do.run(cmd, "GATK pre-alignment {0}".format(region), data)
broad_runner.run_fn("picard_index", pa_bam)
dbsnp_vcf = data["genome_resources"]["variation"]["dbsnp"]
recal_file = realign.gatk_realigner_targets(broad_runner, pa_bam, data["sam_ref"],
dbsnp=dbsnp_vcf, region=region_to_gatk(region))
recal_cl = realign.gatk_indel_realignment_cl(broad_runner, pa_bam, data["sam_ref"],
recal_file, tmp_dir, region=region_to_gatk(region))
return pa_bam, " ".join(recal_cl)