def _piped_input_cl(data, region, tmp_dir, out_base_file, prep_params):
"""Retrieve the commandline for streaming input into preparation step.
If marking duplicates, this requires writing an intermediate file since
MarkDuplicates uses multiple passed on an input.
"""
broad_runner = broad.runner_from_config(data["config"])
cl = _gatk_extract_reads_cl(data, region, prep_params, tmp_dir)
if prep_params["dup"] == "picard":
sel_file = "%s-select%s" % os.path.splitext(out_base_file)
if not utils.file_exists(sel_file):
with file_transaction(sel_file) as tx_out_file:
cl += ["-o", tx_out_file]
do.run_memory_retry(cl,
"GATK: PrintReads",
data,
region=region)
dup_metrics = "%s-dup.dup_metrics" % os.path.splitext(out_base_file)[0]
compression = "5" if prep_params["realign"] == "gatk" else "0"
cl = broad_runner.cl_picard("MarkDuplicates",
[("INPUT", sel_file),
("OUTPUT", "/dev/stdout"),
("METRICS_FILE", dup_metrics),
("PROGRAM_RECORD_ID", "null"),
("COMPRESSION_LEVEL", compression),
("TMP_DIR", tmp_dir)])
elif not prep_params["dup"]:
sel_file = data["work_bam"]
else:
raise ValueError("Duplication approach not supported with GATK: %s" %
prep_params["dup"])
bam.index(sel_file, data["config"])
return sel_file, " ".join(cl)
def _piped_input_cl(data, region, tmp_dir, out_base_file, prep_params):
"""Retrieve the commandline for streaming input into preparation step.
If marking duplicates, this requires writing an intermediate file since
MarkDuplicates uses multiple passed on an input.
"""
broad_runner = broad.runner_from_config(data["config"])
cl = _gatk_extract_reads_cl(data, region, prep_params, tmp_dir)
if prep_params["dup"] == "picard":
sel_file = "%s-select%s" % os.path.splitext(out_base_file)
if not utils.file_exists(sel_file):
with file_transaction(sel_file) as tx_out_file:
cl += ["-o", tx_out_file]
do.run_memory_retry(cl, "GATK: PrintReads", data, region=region)
dup_metrics = "%s-dup.dup_metrics" % os.path.splitext(out_base_file)[0]
compression = "5" if prep_params["realign"] == "gatk" else "0"
cl = broad_runner.cl_picard("MarkDuplicates",
[("INPUT", sel_file),
("OUTPUT", "/dev/stdout"),
("METRICS_FILE", dup_metrics),
("PROGRAM_RECORD_ID", "null"),
("COMPRESSION_LEVEL", compression),
("TMP_DIR", tmp_dir)])
elif not prep_params["dup"]:
sel_file = data["work_bam"]
else:
raise ValueError("Duplication approach not supported with GATK: %s" % prep_params["dup"])
bam.index(sel_file, data["config"])
return sel_file, " ".join(cl)
def run_gatk(self,
params,
tmp_dir=None,
log_error=True,
memory_retry=False,
data=None,
region=None):
with curdir_tmpdir() as local_tmp_dir:
if tmp_dir is None:
tmp_dir = local_tmp_dir
cl = self.cl_gatk(params, tmp_dir)
atype_index = cl.index("-T") if cl.count("-T") > 0 \
else cl.index("--analysis_type")
prog = cl[atype_index + 1]
if memory_retry:
do.run_memory_retry(cl,
"GATK: {0}".format(prog),
data,
region=region)
else:
do.run(cl,
"GATK: {0}".format(prog),
data,
region=region,
log_error=log_error)
def run_gatk(self, params, tmp_dir=None, log_error=True, memory_retry=False, data=None, region=None):
with curdir_tmpdir() as local_tmp_dir:
if tmp_dir is None:
tmp_dir = local_tmp_dir
cl = self.cl_gatk(params, tmp_dir)
atype_index = cl.index("-T") if cl.count("-T") > 0 else cl.index("--analysis_type")
prog = cl[atype_index + 1]
if memory_retry:
do.run_memory_retry(cl, "GATK: {0}".format(prog), data, region=region)
else:
do.run(cl, "GATK: {0}".format(prog), data, region=region, log_error=log_error)
def _piped_bamprep_region_fullpipe(data, region, prep_params, out_file, tmp_dir):
"""Perform fully piped BAM preparation using non-GATK/Picard tools.
"""
with file_transaction(out_file) as tx_out_file:
extract_recal_cmd = _piped_extract_recal_cmd(data, region, prep_params, tmp_dir)
dedup_cmd = _piped_dedup_recal_cmd(data, prep_params, tmp_dir, out_file)
realign_cmd = _piped_realign_cmd(data, prep_params, tmp_dir)
cmd = "{extract_recal_cmd} {dedup_cmd} {realign_cmd} > {tx_out_file}"
cmd = cmd.format(**locals())
do.run_memory_retry(cmd, "Piped post-alignment bamprep {0}".format(region), data,
region=region)
def _piped_bamprep_region_fullpipe(data, region, prep_params, out_file,
tmp_dir):
"""Perform fully piped BAM preparation using non-GATK/Picard tools.
"""
with file_transaction(out_file) as tx_out_file:
extract_recal_cmd = _piped_extract_recal_cmd(data, region, prep_params,
tmp_dir)
dedup_cmd = _piped_dedup_recal_cmd(data, prep_params, tmp_dir,
out_file)
realign_cmd = _piped_realign_cmd(data, prep_params, tmp_dir)
cmd = "{extract_recal_cmd} {dedup_cmd} {realign_cmd} > {tx_out_file}"
cmd = cmd.format(**locals())
do.run_memory_retry(cmd,
"Piped post-alignment bamprep {0}".format(region),
data,
region=region)