def _convert_bam_to_fastq(in_file, work_dir, data, dirs, config):
"""Convert BAM input file into FASTQ files.
"""
out_dir = safe_makedir(os.path.join(work_dir, "fastq_convert"))
qual_bin_method = config["algorithm"].get("quality_bin")
if (qual_bin_method == "prealignment"
or (isinstance(qual_bin_method, list)
and "prealignment" in qual_bin_method)):
out_bindir = safe_makedir(os.path.join(out_dir, "qualbin"))
in_file = cram.illumina_qual_bin(in_file, data["sam_ref"], out_bindir,
config)
out_files = [
os.path.join(
out_dir, "{0}_{1}.fastq".format(
os.path.splitext(os.path.basename(in_file))[0], x))
for x in ["1", "2"]
]
if bam.is_paired(in_file):
out1, out2 = out_files
else:
out1 = out_files[0]
out2 = None
if not file_exists(out1):
broad_runner = broad.runner_from_config(config)
broad_runner.run_fn("picard_bam_to_fastq", in_file, out1, out2)
if out2 and os.path.getsize(out2) == 0:
out2 = None
return [out1, out2]
def _convert_bam_to_fastq(in_file, work_dir, data, dirs, config):
"""Convert BAM input file into FASTQ files.
"""
out_dir = safe_makedir(os.path.join(work_dir, "fastq_convert"))
qual_bin_method = config["algorithm"].get("quality_bin")
if (qual_bin_method == "prealignment" or
(isinstance(qual_bin_method, list) and "prealignment" in qual_bin_method)):
out_bindir = safe_makedir(os.path.join(out_dir, "qualbin"))
in_file = cram.illumina_qual_bin(in_file, data["sam_ref"], out_bindir, config)
out_files = [os.path.join(out_dir, "{0}_{1}.fastq".format(
os.path.splitext(os.path.basename(in_file))[0], x))
for x in ["1", "2"]]
if bam.is_paired(in_file):
out1, out2 = out_files
else:
out1 = out_files[0]
out2 = None
if not file_exists(out1):
broad_runner = broad.runner_from_path("picard", config)
broad_runner.run_fn("picard_bam_to_fastq", in_file, out1, out2)
if out2 and os.path.getsize(out2) == 0:
out2 = None
return [out1, out2]
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 _align_from_bam(fastq1, aligner, align_ref, sam_ref, names, align_dir, config):
qual_bin_method = config["algorithm"].get("quality_bin")
if qual_bin_method == "prealignment" or (isinstance(qual_bin_method, list) and "prealignment" in qual_bin_method):
out_dir = utils.safe_makedir(os.path.join(align_dir, "qualbin"))
fastq1 = cram.illumina_qual_bin(fastq1, sam_ref, out_dir, config)
align_fn = _tools[aligner].bam_align_fn
return align_fn(fastq1, align_ref, names, align_dir, config)
def _align_from_bam(fastq1, aligner, align_ref, sam_ref, names, align_dir, config):
qual_bin_method = config["algorithm"].get("quality_bin")
if (qual_bin_method == "prealignment" or
(isinstance(qual_bin_method, list) and "prealignment" in qual_bin_method)):
out_dir = utils.safe_makedir(os.path.join(align_dir, "qualbin"))
fastq1 = cram.illumina_qual_bin(fastq1, sam_ref, out_dir, config)
align_fn = _tools[aligner].bam_align_fn
if align_fn is None:
raise NotImplementedError("Do not yet support BAM alignment with %s" % aligner)
return align_fn(fastq1, align_ref, names, align_dir, config)
def _align_from_bam(fastq1, aligner, align_ref, sam_ref, names, align_dir, data):
assert not data.get("align_split"), "Do not handle split alignments with BAM yet"
config = data["config"]
qual_bin_method = config["algorithm"].get("quality_bin")
if qual_bin_method == "prealignment" or (isinstance(qual_bin_method, list) and "prealignment" in qual_bin_method):
out_dir = utils.safe_makedir(os.path.join(align_dir, "qualbin"))
fastq1 = cram.illumina_qual_bin(fastq1, sam_ref, out_dir, config)
align_fn = TOOLS[aligner].bam_align_fn
if align_fn is None:
raise NotImplementedError("Do not yet support BAM alignment with %s" % aligner)
return align_fn(fastq1, align_ref, names, align_dir, config)
def split_read_files(fastq1, fastq2, item, split_size, out_dir, dirs, config):
"""Split input reads for parallel processing, dispatching on input type.
"""
if fastq1.endswith(".bam") and fastq2 is None:
qual_bin_method = config["algorithm"].get("quality_bin")
if (qual_bin_method == "prealignment" or
(isinstance(qual_bin_method, list) and "prealignment" in qual_bin_method)):
out_bindir = utils.safe_makedir(os.path.join(out_dir, "qualbin"))
fastq1 = cram.illumina_qual_bin(fastq1, item["sam_ref"], out_bindir, config)
return split_bam_file(fastq1, split_size, out_dir, config)
else:
return split_fastq_files(fastq1, fastq2, split_size, out_dir, config)
def _align_from_bam(fastq1, aligner, align_ref, sam_ref, names, align_dir,
config):
qual_bin_method = config["algorithm"].get("quality_bin")
if (qual_bin_method == "prealignment"
or (isinstance(qual_bin_method, list)
and "prealignment" in qual_bin_method)):
out_dir = utils.safe_makedir(os.path.join(align_dir, "qualbin"))
fastq1 = cram.illumina_qual_bin(fastq1, sam_ref, out_dir, config)
align_fn = TOOLS[aligner].bam_align_fn
if align_fn is None:
raise NotImplementedError("Do not yet support BAM alignment with %s" %
aligner)
return align_fn(fastq1, align_ref, names, align_dir, config)
def _align_from_bam(fastq1, aligner, align_ref, sam_ref, names, align_dir, data):
assert not data.get("align_split"), "Do not handle split alignments with BAM yet"
config = data["config"]
qual_bin_method = config["algorithm"].get("quality_bin")
if (qual_bin_method == "prealignment" or
(isinstance(qual_bin_method, list) and "prealignment" in qual_bin_method)):
out_dir = utils.safe_makedir(os.path.join(align_dir, "qualbin"))
fastq1 = cram.illumina_qual_bin(fastq1, sam_ref, out_dir, config)
align_fn = TOOLS[aligner].bam_align_fn
if align_fn is None:
raise NotImplementedError("Do not yet support BAM alignment with %s" % aligner)
out = align_fn(fastq1, align_ref, names, align_dir, data)
if isinstance(out, dict):
assert "work_bam" in out
return out
else:
data["work_bam"] = out
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]