def piped_bamprep(data, region=None, out_file=None):
"""Perform full BAM preparation using pipes to avoid intermediate disk IO.
Handles recalibration and realignment of original BAMs.
"""
data["region"] = region
if not _need_prep(data):
return [data]
else:
utils.safe_makedir(os.path.dirname(out_file))
if region[0] == "nochrom":
prep_bam = shared.write_nochr_reads(data["work_bam"], out_file,
data["config"])
elif region[0] == "noanalysis":
prep_bam = shared.write_noanalysis_reads(data["work_bam"],
region[1], out_file,
data["config"])
else:
if not utils.file_exists(out_file):
with tx_tmpdir(data) as tmp_dir:
_piped_bamprep_region(data, region, out_file, tmp_dir)
prep_bam = out_file
bam.index(prep_bam, data["config"])
data["work_bam"] = prep_bam
return [data]
def write_recal_bam(data, region=None, out_file=None):
"""Step 2 of GATK recalibration -- use covariates to re-write output file.
"""
config = data["config"]
if out_file is None:
out_file = "%s-gatkrecal.bam" % os.path.splitext(data["work_bam"])[0]
logger.info("Writing recalibrated BAM for %s to %s" %
(data["name"], out_file))
if region == "nochr":
out_bam = write_nochr_reads(data["work_bam"], out_file)
else:
out_bam = _run_recal_bam(data["work_bam"], data["prep_recal"], region,
data["sam_ref"], out_file, config)
qual_bin = config["algorithm"].get("quality_bin", None)
if ((qual_bin is True or qual_bin == "postrecal"
or isinstance(qual_bin, list) and "postrecal" in qual_bin)
and has_aligned_reads(out_bam)):
binned_bam = cram.illumina_qual_bin(out_bam, data["sam_ref"],
os.path.dirname(out_bam), config)
shutil.move(out_bam, out_bam + ".binned")
shutil.move(binned_bam, out_bam)
utils.save_diskspace(out_bam + ".binned",
"Quality binned to %s" % out_bam, config)
data["work_bam"] = out_bam
return [data]
def realign_sample(data, region=None, out_file=None):
"""Realign sample BAM file at indels.
"""
realigner = data["config"]["algorithm"].get("realign", True)
realigner = "gatk" if realigner is True else realigner
realign_fn = _realign_approaches[realigner] if realigner else None
if realign_fn:
logger.info("Realigning %s with %s: %s %s" %
(data["name"], realigner, os.path.basename(
data["work_bam"]), region))
sam_ref = data["sam_ref"]
config = data["config"]
if region == "nochr":
realign_bam = write_nochr_reads(data["work_bam"], out_file,
data["config"])
else:
realign_bam = realign_fn(
data["work_bam"], sam_ref, config,
data["genome_resources"]["variation"]["dbsnp"], region,
out_file)
if region is None:
save_diskspace(data["work_bam"], "Realigned to %s" % realign_bam,
config)
data["work_bam"] = realign_bam
return [data]
def write_recal_bam(data, region=None, out_file=None):
"""Step 2 of GATK recalibration -- use covariates to re-write output file.
"""
config = data["config"]
if out_file is None:
out_file = "%s-gatkrecal.bam" % os.path.splitext(data["work_bam"])[0]
logger.info("Writing recalibrated BAM for %s to %s" % (data["name"], out_file))
if region == "nochr":
out_bam = write_nochr_reads(data["work_bam"], out_file)
else:
out_bam = _run_recal_bam(data["work_bam"], data["prep_recal"],
region, data["sam_ref"], out_file, config)
data["work_bam"] = out_bam
return [data]
def write_recal_bam(data, region=None, out_file=None):
"""Step 2 of GATK recalibration -- use covariates to re-write output file.
"""
config = data["config"]
if out_file is None:
out_file = "%s-gatkrecal.bam" % os.path.splitext(data["work_bam"])[0]
logger.info("Writing recalibrated BAM for %s to %s" %
(data["name"], out_file))
if region == "nochr":
out_bam = write_nochr_reads(data["work_bam"], out_file)
else:
out_bam = _run_recal_bam(data["work_bam"], data["prep_recal"], region,
data["sam_ref"], out_file, config)
data["work_bam"] = out_bam
return [data]
def realign_sample(data, region=None, out_file=None):
"""Realign sample BAM file at indels.
"""
logger.info("Realigning %s with GATK: %s %s" % (data["name"], os.path.basename(data["work_bam"]), region))
if data["config"]["algorithm"]["snpcall"]:
sam_ref = data["sam_ref"]
config = data["config"]
if region == "nochr":
realign_bam = write_nochr_reads(data["work_bam"], out_file)
else:
realign_bam = gatk_realigner(
data["work_bam"], sam_ref, config, configured_ref_file("dbsnp", config, sam_ref), region, out_file
)
if region is None:
save_diskspace(data["work_bam"], "Realigned to %s" % realign_bam, config)
data["work_bam"] = realign_bam
return [data]
def piped_bamprep(data, region=None, out_file=None):
"""Perform full BAM preparation using pipes to avoid intermediate disk IO.
Handles de-duplication, recalibration and realignment of original BAMs.
"""
if region[0] == "nochrom":
prep_bam = shared.write_nochr_reads(data["work_bam"], out_file)
elif region[0] == "noanalysis":
prep_bam = shared.write_noanalysis_reads(data["work_bam"], region[1], out_file)
else:
if not utils.file_exists(out_file):
with utils.curdir_tmpdir() as tmp_dir:
_piped_bamprep_region(data, region, out_file, tmp_dir)
prep_bam = out_file
data["work_bam"] = prep_bam
data["regions"]["current"] = region
return [data]
def piped_bamprep(data, region=None, out_file=None):
"""Perform full BAM preparation using pipes to avoid intermediate disk IO.
Handles de-duplication, recalibration and realignment of original BAMs.
"""
utils.safe_makedir(os.path.dirname(out_file))
if region[0] == "nochrom":
prep_bam = shared.write_nochr_reads(data["work_bam"], out_file)
elif region[0] == "noanalysis":
prep_bam = shared.write_noanalysis_reads(data["work_bam"], region[1], out_file)
else:
if not utils.file_exists(out_file):
with utils.curdir_tmpdir() as tmp_dir:
_piped_bamprep_region(data, region, out_file, tmp_dir)
prep_bam = out_file
broad_runner = broad.runner_from_config(data["config"])
broad_runner.run_fn("picard_index", prep_bam)
data["work_bam"] = prep_bam
data["region"] = region
return [data]
def realign_sample(data, region=None, out_file=None):
"""Realign sample BAM file at indels.
"""
logger.info("Realigning %s with GATK: %s %s" %
(data["name"], os.path.basename(data["work_bam"]), region))
if data["config"]["algorithm"]["snpcall"]:
sam_ref = data["sam_ref"]
config = data["config"]
if region == "nochr":
realign_bam = write_nochr_reads(data["work_bam"], out_file)
else:
realign_bam = gatk_realigner(
data["work_bam"], sam_ref, config,
configured_ref_file("dbsnp", config, sam_ref), region,
out_file)
if region is None:
save_diskspace(data["work_bam"], "Realigned to %s" % realign_bam,
config)
data["work_bam"] = realign_bam
return [data]
def piped_bamprep(data, region=None, out_file=None):
"""Perform full BAM preparation using pipes to avoid intermediate disk IO.
Handles de-duplication, recalibration and realignment of original BAMs.
"""
utils.safe_makedir(os.path.dirname(out_file))
if region[0] == "nochrom":
prep_bam = shared.write_nochr_reads(data["work_bam"], out_file)
elif region[0] == "noanalysis":
prep_bam = shared.write_noanalysis_reads(data["work_bam"], region[1], out_file)
else:
if not utils.file_exists(out_file):
with utils.curdir_tmpdir() as tmp_dir:
_piped_bamprep_region(data, region, out_file, tmp_dir)
prep_bam = out_file
broad_runner = broad.runner_from_config(data["config"])
broad_runner.run_fn("picard_index", prep_bam)
data["work_bam"] = prep_bam
data["region"] = region
return [data]
def piped_bamprep(data, region=None, out_file=None):
"""Perform full BAM preparation using pipes to avoid intermediate disk IO.
Handles recalibration and realignment of original BAMs.
"""
data["region"] = region
if not _need_prep(data):
return [data]
else:
utils.safe_makedir(os.path.dirname(out_file))
if region[0] == "nochrom":
prep_bam = shared.write_nochr_reads(data["work_bam"], out_file, data["config"])
elif region[0] == "noanalysis":
prep_bam = shared.write_noanalysis_reads(data["work_bam"], region[1], out_file, data["config"])
else:
if not utils.file_exists(out_file):
with tx_tmpdir(data) as tmp_dir:
_piped_bamprep_region(data, region, out_file, tmp_dir)
prep_bam = out_file
bam.index(prep_bam, data["config"])
data["work_bam"] = prep_bam
return [data]
def realign_sample(data, region=None, out_file=None):
"""Realign sample BAM file at indels.
"""
realigner = data["config"]["algorithm"].get("realign", True)
realigner = "gatk" if realigner is True else realigner
realign_fn = _realign_approaches[realigner] if realigner else None
if realign_fn:
logger.info("Realigning %s with %s: %s %s" % (data["name"], realigner,
os.path.basename(data["work_bam"]),
region))
sam_ref = data["sam_ref"]
config = data["config"]
if region == "nochr":
realign_bam = write_nochr_reads(data["work_bam"], out_file)
else:
realign_bam = realign_fn(data["work_bam"], sam_ref, config,
data["genome_resources"]["variation"]["dbsnp"],
region, out_file)
if region is None:
save_diskspace(data["work_bam"], "Realigned to %s" % realign_bam,
config)
data["work_bam"] = realign_bam
return [data]
def write_recal_bam(data, region=None, out_file=None):
"""Step 2 of GATK recalibration -- use covariates to re-write output file.
"""
config = data["config"]
if out_file is None:
out_file = "%s-gatkrecal.bam" % os.path.splitext(data["work_bam"])[0]
logger.info("Writing recalibrated BAM for %s to %s" % (data["name"], out_file))
if region == "nochr":
out_bam = write_nochr_reads(data["work_bam"], out_file, data["config"])
else:
out_bam = _run_recal_bam(data["work_bam"], data["prep_recal"],
region, data["sam_ref"], out_file, config)
qual_bin = config["algorithm"].get("quality_bin", None)
if ((qual_bin is True or qual_bin == "postrecal" or
isinstance(qual_bin, list) and "postrecal" in qual_bin)
and has_aligned_reads(out_bam)):
binned_bam = cram.illumina_qual_bin(out_bam, data["sam_ref"],
os.path.dirname(out_bam), config)
shutil.move(out_bam, out_bam + ".binned")
shutil.move(binned_bam, out_bam)
utils.save_diskspace(out_bam + ".binned",
"Quality binned to %s" % out_bam, config)
data["work_bam"] = out_bam
return [data]