def _run_recal_bam(dup_align_bam, recal_file, region, ref_file, out_file, config):
"""Run BAM recalibration with the given input
"""
if not file_exists(out_file):
if _recal_available(recal_file):
broad_runner = broad.runner_from_config(config)
with curdir_tmpdir() as tmp_dir:
with file_transaction(out_file) as tx_out_file:
params = ["-T", "PrintReads",
"-BQSR", recal_file,
"-R", ref_file,
"-I", dup_align_bam,
"--out", tx_out_file,
]
base_bed = config["algorithm"].get("variant_regions", None)
region_bed = subset_variant_regions(base_bed, region, tx_out_file)
if region_bed:
params += ["-L", region_bed, "--interval_set_rule", "INTERSECTION"]
elif region:
params += ["-L", region, "--interval_set_rule", "INTERSECTION"]
broad_runner.run_gatk(params, tmp_dir)
elif region:
subset_bam_by_region(dup_align_bam, region, out_file)
else:
shutil.copy(dup_align_bam, out_file)
return out_file
def _run_recal_bam(dup_align_bam, recal_file, region, ref_file, out_file,
config):
"""Run BAM recalibration with the given input
"""
if not file_exists(out_file):
if _recal_available(recal_file):
broad_runner = broad.runner_from_config(config)
intervals = config["algorithm"].get("variant_regions", None)
with curdir_tmpdir() as tmp_dir:
with file_transaction(out_file) as tx_out_file:
params = [
"-T",
"PrintReads",
"-BQSR",
recal_file,
"-R",
ref_file,
"-I",
dup_align_bam,
"--out",
tx_out_file,
]
if region:
params += ["-L", region]
if intervals:
params += ["-L", intervals]
if params and intervals:
params += ["--interval_set_rule", "INTERSECTION"]
broad_runner.run_gatk(params, tmp_dir)
elif region:
subset_bam_by_region(dup_align_bam, region, out_file)
else:
shutil.copy(dup_align_bam, out_file)
return out_file
def gatk_realigner(align_bam, ref_file, config, dbsnp=None, region=None,
out_file=None, deep_coverage=False):
"""Realign a BAM file around indels using GATK, returning sorted BAM.
"""
runner = broad.runner_from_config(config)
runner.run_fn("picard_index", align_bam)
runner.run_fn("picard_index_ref", ref_file)
if not os.path.exists("%s.fai" % ref_file):
pysam.faidx(ref_file)
if region:
align_bam = subset_bam_by_region(align_bam, region, out_file)
runner.run_fn("picard_index", align_bam)
if has_aligned_reads(align_bam, region):
variant_regions = config["algorithm"].get("variant_regions", None)
realign_target_file = gatk_realigner_targets(runner, align_bam,
ref_file, dbsnp, region,
out_file, deep_coverage,
variant_regions)
realign_bam = gatk_indel_realignment(runner, align_bam, ref_file,
realign_target_file, region,
out_file, deep_coverage)
# No longer required in recent GATK (> Feb 2011) -- now done on the fly
# realign_sort_bam = runner.run_fn("picard_fixmate", realign_bam)
return realign_bam
elif out_file:
shutil.copy(align_bam, out_file)
return out_file
else:
return align_bam
def gatk_realigner(align_bam,
ref_file,
config,
dbsnp=None,
region=None,
out_file=None,
deep_coverage=False):
"""Realign a BAM file around indels using GATK, returning sorted BAM.
"""
runner = broad.runner_from_config(config)
runner.run_fn("picard_index", align_bam)
runner.run_fn("picard_index_ref", ref_file)
if not os.path.exists("%s.fai" % ref_file):
pysam.faidx(ref_file)
if region:
align_bam = subset_bam_by_region(align_bam, region, out_file)
runner.run_fn("picard_index", align_bam)
if has_aligned_reads(align_bam, region):
variant_regions = config["algorithm"].get("variant_regions", None)
realign_target_file = gatk_realigner_targets(runner, align_bam,
ref_file, dbsnp, region,
out_file, deep_coverage,
variant_regions)
realign_bam = gatk_indel_realignment(runner, align_bam, ref_file,
realign_target_file, region,
out_file, deep_coverage)
# No longer required in recent GATK (> Feb 2011) -- now done on the fly
# realign_sort_bam = runner.run_fn("picard_fixmate", realign_bam)
return realign_bam
elif out_file:
shutil.copy(align_bam, out_file)
return out_file
else:
return align_bam