def samblaster_dedup_sort(data, tx_out_file, tx_sr_file, tx_disc_file):
"""Deduplicate and sort with samblaster, produces split read and discordant pair files.
"""
samblaster = config_utils.get_program("samblaster", data["config"])
samtools = config_utils.get_program("samtools", data["config"])
tmp_prefix = "%s-sorttmp" % utils.splitext_plus(tx_out_file)[0]
tobam_cmd = ("{samtools} sort {sort_opt} -@ {cores} -m {mem} -T {tmp_prefix}-{dext} {out_file} -")
# full BAM -- associate more memory and cores
cores, mem = _get_cores_memory(data, downscale=2)
# Potentially downsample to maximum coverage here if not splitting and whole genome sample
ds_cmd = None if data.get("align_split") else bam.get_maxcov_downsample_cl(data, "samtools")
sort_opt = "-n" if data.get("align_split") and dd.get_mark_duplicates(data) else ""
if ds_cmd:
dedup_cmd = "%s %s > %s" % (tobam_cmd.format(out_file="", dext="full", **locals()), ds_cmd, tx_out_file)
else:
dedup_cmd = tobam_cmd.format(out_file="-o %s" % tx_out_file, dext="full", **locals())
# split and discordant BAMs -- give less memory/cores since smaller files
sort_opt = ""
cores, mem = _get_cores_memory(data, downscale=4)
splitter_cmd = tobam_cmd.format(out_file="-o %s" % tx_sr_file, dext="spl", **locals())
discordant_cmd = tobam_cmd.format(out_file="-o %s" % tx_disc_file, dext="disc", **locals())
# samblaster 0.1.22 and better require the -M flag for compatibility with bwa-mem
cmd = ("{samblaster} --addMateTags -M --splitterFile >({splitter_cmd}) --discordantFile >({discordant_cmd}) "
"| {dedup_cmd}")
return cmd.format(**locals())
def align_to_sort_bam(fastq1, fastq2, aligner, data):
"""Align to the named genome build, returning a sorted BAM file.
"""
names = data["rgnames"]
align_dir_parts = [data["dirs"]["work"], "align", names["sample"]]
if data.get("disambiguate"):
align_dir_parts.append(data["disambiguate"]["genome_build"])
aligner_index = _get_aligner_index(aligner, data)
align_dir = utils.safe_makedir(apply(os.path.join, align_dir_parts))
ref_file = tz.get_in(("reference", "fasta", "base"), data)
if fastq1.endswith(".bam"):
data = _align_from_bam(fastq1, aligner, aligner_index, ref_file, names,
align_dir, data)
else:
data = _align_from_fastq(fastq1, fastq2, aligner, aligner_index,
ref_file, names, align_dir, data)
if data["work_bam"] and utils.file_exists(data["work_bam"]):
if data.get("align_split") and dd.get_mark_duplicates(data):
# If merging later with with bamsormadup need query sorted inputs
# but CWL requires a bai file. Create a fake one to make it happy.
bam.fake_index(data["work_bam"], data)
else:
bam.index(data["work_bam"], data["config"])
for extra in ["-sr", "-disc"]:
extra_bam = utils.append_stem(data['work_bam'], extra)
if utils.file_exists(extra_bam):
bam.index(extra_bam, data["config"])
return data
def clean_chipseq_alignment(data):
aligner = dd.get_aligner(data)
data["align_bam"] = dd.get_work_bam(data)
if dd.get_mark_duplicates(data):
if aligner:
if aligner == "bowtie2":
filterer = bowtie2.filter_multimappers
elif aligner == "bwa":
filterer = bwa.filter_multimappers
else:
logger.error("ChIP-seq only supported for bowtie2 and bwa.")
sys.exit(-1)
unique_bam = filterer(dd.get_work_bam(data), data)
data["work_bam"] = unique_bam
else:
logger.info(
"Warning: When BAM file is given as input, bcbio skips multimappers removal."
"If BAM is not cleaned for peak calling, can result in downstream errors."
)
# lcr_bed = utils.get_in(data, ("genome_resources", "variation", "lcr"))
data["work_bam"] = _keep_assembled_chrom(dd.get_work_bam(data),
dd.get_ref_file(data),
data["config"])
encode_bed = tz.get_in(
["genome_resources", "variation", "encode_blacklist"], data)
if encode_bed:
data["work_bam"] = _prepare_bam(dd.get_work_bam(data), encode_bed,
data['config'])
bam.index(data["work_bam"], data['config'])
data["bigwig"] = _bam_coverage(dd.get_sample_name(data),
dd.get_work_bam(data), data)
return [[data]]
def clean_chipseq_alignment(data):
aligner = dd.get_aligner(data)
data["align_bam"] = dd.get_work_bam(data)
if dd.get_mark_duplicates(data):
if aligner:
if aligner == "bowtie2":
filterer = bowtie2.filter_multimappers
elif aligner == "bwa":
filterer = bwa.filter_multimappers
else:
logger.error("ChIP-seq only supported for bowtie2 and bwa.")
sys.exit(-1)
unique_bam = filterer(dd.get_work_bam(data), data)
data["work_bam"] = unique_bam
else:
logger.info("Warning: When BAM file is given as input, bcbio skips multimappers removal."
"If BAM is not cleaned for peak calling, can result in downstream errors.")
# lcr_bed = utils.get_in(data, ("genome_resources", "variation", "lcr"))
data["work_bam"] = _keep_assembled_chrom(dd.get_work_bam(data), dd.get_ref_file(data),
data["config"])
encode_bed = tz.get_in(["genome_resources", "variation", "encode_blacklist"], data)
if encode_bed:
data["work_bam"] = _prepare_bam(dd.get_work_bam(data), encode_bed, data['config'])
bam.index(data["work_bam"], data['config'])
data["bigwig"] = _bam_coverage(dd.get_sample_name(data), dd.get_work_bam(data), data)
return [[data]]
def samblaster_dedup_sort(data, tx_out_file, tx_sr_file, tx_disc_file):
"""Deduplicate and sort with samblaster, produces split read and discordant pair files.
"""
samblaster = config_utils.get_program("samblaster", data["config"])
samtools = config_utils.get_program("samtools", data["config"])
tmp_prefix = "%s-sorttmp" % utils.splitext_plus(tx_out_file)[0]
tobam_cmd = ("{samtools} sort {sort_opt} -@ {cores} -m {mem} -T {tmp_prefix}-{dext} {out_file} -")
# full BAM -- associate more memory and cores
cores, mem = _get_cores_memory(data, downscale=2)
# Potentially downsample to maximum coverage here if not splitting and whole genome sample
ds_cmd = None if data.get("align_split") else bam.get_maxcov_downsample_cl(data, "samtools")
sort_opt = "-n" if data.get("align_split") and dd.get_mark_duplicates(data) else ""
if ds_cmd:
dedup_cmd = "%s %s > %s" % (tobam_cmd.format(out_file="", dext="full", **locals()), ds_cmd, tx_out_file)
else:
dedup_cmd = tobam_cmd.format(out_file="-o %s" % tx_out_file, dext="full", **locals())
# split and discordant BAMs -- give less memory/cores since smaller files
sort_opt = ""
cores, mem = _get_cores_memory(data, downscale=4)
splitter_cmd = tobam_cmd.format(out_file="-o %s" % tx_sr_file, dext="spl", **locals())
discordant_cmd = tobam_cmd.format(out_file="-o %s" % tx_disc_file, dext="disc", **locals())
# samblaster 0.1.22 and better require the -M flag for compatibility with bwa-mem
cmd = ("{samblaster} --addMateTags -M --splitterFile >({splitter_cmd}) --discordantFile >({discordant_cmd}) "
"| {dedup_cmd}")
return cmd.format(**locals())
def align_to_sort_bam(fastq1, fastq2, aligner, data):
"""Align to the named genome build, returning a sorted BAM file.
"""
names = data["rgnames"]
align_dir_parts = [data["dirs"]["work"], "align", names["sample"]]
if data.get("disambiguate"):
align_dir_parts.append(data["disambiguate"]["genome_build"])
aligner_index = _get_aligner_index(aligner, data)
align_dir = utils.safe_makedir(apply(os.path.join, align_dir_parts))
ref_file = tz.get_in(("reference", "fasta", "base"), data)
if fastq1.endswith(".bam"):
data = _align_from_bam(fastq1, aligner, aligner_index, ref_file,
names, align_dir, data)
else:
data = _align_from_fastq(fastq1, fastq2, aligner, aligner_index, ref_file,
names, align_dir, data)
if data["work_bam"] and utils.file_exists(data["work_bam"]):
if data.get("align_split") and dd.get_mark_duplicates(data):
# If merging later with with bamsormadup need query sorted inputs
# but CWL requires a bai file. Create a fake one to make it happy.
bam.fake_index(data["work_bam"], data)
else:
bam.index(data["work_bam"], data["config"])
for extra in ["-sr", "-disc"]:
extra_bam = utils.append_stem(data['work_bam'], extra)
if utils.file_exists(extra_bam):
bam.index(extra_bam, data["config"])
return data
def merge_bam_files(bam_files, work_dir, data, out_file=None, batch=None):
"""Merge multiple BAM files from a sample into a single BAM for processing.
Checks system open file limit and merges in batches if necessary to avoid
file handle limits.
"""
out_file = _merge_outfile_fname(out_file, bam_files, work_dir, batch)
if not utils.file_exists(out_file):
if len(bam_files) == 1 and bam.bam_already_sorted(
bam_files[0], data["config"], "coordinate"):
with file_transaction(data, out_file) as tx_out_file:
_create_merge_filelist(bam_files, tx_out_file, data["config"])
out_file = bam_files[0]
samtools = config_utils.get_program("samtools", data["config"])
do.run('{} quickcheck -v {}'.format(samtools, out_file),
"Check for valid merged BAM after transfer")
else:
with tx_tmpdir(data) as tmpdir:
with utils.chdir(tmpdir):
with file_transaction(data, out_file) as tx_out_file:
tx_bam_file_list = _create_merge_filelist(
bam_files, tx_out_file, data["config"])
sambamba = config_utils.get_program(
"sambamba", data["config"])
samtools = config_utils.get_program(
"samtools", data["config"])
resources = config_utils.get_resources(
"samtools", data["config"])
num_cores = dd.get_num_cores(data)
# Aim for 3.5Gb/core memory for BAM merging
num_cores = config_utils.adjust_cores_to_mb_target(
3500, resources.get("memory", "2G"), num_cores)
max_mem = config_utils.adjust_memory(
resources.get("memory", "1G"), 2,
"decrease").upper()
if dd.get_mark_duplicates(data):
cmd = _biobambam_merge_dedup_maxcov(data)
else:
cmd = _biobambam_merge_maxcov(data)
do.run(
cmd.format(**locals()), "Merge bam files to %s" %
os.path.basename(out_file), None)
do.run(
'{} quickcheck -v {}'.format(
samtools, tx_out_file),
"Check for valid merged BAM")
do.run('{} quickcheck -v {}'.format(samtools, out_file),
"Check for valid merged BAM after transfer")
_finalize_merge(out_file, bam_files, data["config"])
bam.index(out_file, data["config"])
return out_file
def run(align_bams, items, ref_file, assoc_files, region, out_file):
"""Run platypus variant calling, germline whole genome or exome.
"""
assert out_file.endswith(".vcf.gz")
if not utils.file_exists(out_file):
with file_transaction(items[0], out_file) as tx_out_file:
for align_bam in align_bams:
bam.index(align_bam, items[0]["config"])
cmd = [
"platypus", "callVariants",
"--regions=%s" % _subset_regions(region, out_file, items),
"--bamFiles=%s" % ",".join(align_bams),
"--refFile=%s" % dd.get_ref_file(items[0]), "--output=-",
"--logFileName", "/dev/null", "--verbosity=1"
]
resources = config_utils.get_resources("platypus",
items[0]["config"])
if resources.get("options"):
# normalize options so we can set defaults without overwriting user specified
for opt in resources["options"]:
if "=" in opt:
key, val = opt.split("=")
cmd.extend([key, val])
else:
cmd.append(opt)
if any("gvcf" in dd.get_tools_on(d) for d in items):
cmd += ["--outputRefCalls", "1", "--refCallBlockSize", "50000"]
# Adjust default filter thresholds to achieve similar sensitivity/specificity to other callers
# Currently not used after doing more cross validation as they increase false positives
# which seems to be a major advantage for Platypus users.
# tuned_opts = ["--hapScoreThreshold", "10", "--scThreshold", "0.99", "--filteredReadsFrac", "0.9",
# "--rmsmqThreshold", "20", "--qdThreshold", "0", "--abThreshold", "0.0001",
# "--minVarFreq", "0.0", "--assemble", "1"]
# for okey, oval in utils.partition_all(2, tuned_opts):
# if okey not in cmd:
# cmd.extend([okey, oval])
# Avoid filtering duplicates on high depth targeted regions where we don't mark duplicates
if any(not dd.get_mark_duplicates(data) for data in items):
cmd += ["--filterDuplicates=0"]
post_process_cmd = (
" | %s | %s | %s | vcfallelicprimitives -t DECOMPOSED --keep-geno | vcffixup - | "
"vcfstreamsort | bgzip -c > %s" %
(vcfutils.fix_ambiguous_cl(), vcfutils.fix_ambiguous_cl(5),
vcfutils.add_contig_to_header_cl(dd.get_ref_file(items[0]),
tx_out_file), tx_out_file))
do.run(" ".join(cmd) + post_process_cmd,
"platypus variant calling")
out_file = vcfutils.bgzip_and_index(out_file, items[0]["config"])
return out_file
def _biobambam_dedup_sort(data, tx_out_file):
"""Perform streaming deduplication and sorting with biobambam's bamsormadup
"""
samtools = config_utils.get_program("samtools", data["config"])
cores, mem = _get_cores_memory(data, downscale=2)
tmp_file = "%s-sorttmp" % utils.splitext_plus(tx_out_file)[0]
if data.get("align_split"):
sort_opt = "-n" if data.get("align_split") and dd.get_mark_duplicates(data) else ""
cmd = "{samtools} sort %s -@ {cores} -m {mem} -O bam -T {tmp_file}-namesort -o {tx_out_file} -" % sort_opt
else:
ds_cmd = bam.get_maxcov_downsample_cl(data, "bamsormadup")
cmd = ("bamsormadup inputformat=sam threads={cores} tmpfile={tmp_file}-markdup "
"SO=coordinate %s > {tx_out_file}" % ds_cmd)
return cmd.format(**locals())
def _biobambam_dedup_sort(data, tx_out_file):
"""Perform streaming deduplication and sorting with biobambam's bamsormadup
"""
samtools = config_utils.get_program("samtools", data["config"])
cores, mem = _get_cores_memory(data, downscale=2)
tmp_file = "%s-sorttmp" % utils.splitext_plus(tx_out_file)[0]
if data.get("align_split"):
sort_opt = "-n" if data.get("align_split") and dd.get_mark_duplicates(data) else ""
cmd = "{samtools} sort %s -@ {cores} -m {mem} -O bam -T {tmp_file}-namesort -o {tx_out_file} -" % sort_opt
else:
ds_cmd = bam.get_maxcov_downsample_cl(data, "bamsormadup")
bamsormadup = config_utils.get_program("bamsormadup", data)
cmd = ("{bamsormadup} inputformat=sam threads={cores} tmpfile={tmp_file}-markdup "
"SO=coordinate %s > {tx_out_file}" % ds_cmd)
return cmd.format(**locals())
def run(align_bams, items, ref_file, assoc_files, region, out_file):
"""Run platypus variant calling, germline whole genome or exome.
"""
assert out_file.endswith(".vcf.gz")
if not utils.file_exists(out_file):
with file_transaction(items[0], out_file) as tx_out_file:
for align_bam in align_bams:
bam.index(align_bam, items[0]["config"])
cmd = ["platypus", "callVariants", "--regions=%s" % _subset_regions(region, out_file, items),
"--bamFiles=%s" % ",".join(align_bams),
"--refFile=%s" % dd.get_ref_file(items[0]), "--output=-",
"--logFileName", "/dev/null", "--verbosity=1"]
resources = config_utils.get_resources("platypus", items[0]["config"])
if resources.get("options"):
# normalize options so we can set defaults without overwriting user specified
for opt in resources["options"]:
if "=" in opt:
key, val = opt.split("=")
cmd.extend([key, val])
else:
cmd.append(opt)
if any("gvcf" in dd.get_tools_on(d) for d in items):
cmd += ["--outputRefCalls", "1", "--refCallBlockSize", "50000"]
# Adjust default filter thresholds to achieve similar sensitivity/specificity to other callers
# Currently not used after doing more cross validation as they increase false positives
# which seems to be a major advantage for Platypus users.
# tuned_opts = ["--hapScoreThreshold", "10", "--scThreshold", "0.99", "--filteredReadsFrac", "0.9",
# "--rmsmqThreshold", "20", "--qdThreshold", "0", "--abThreshold", "0.0001",
# "--minVarFreq", "0.0", "--assemble", "1"]
# for okey, oval in utils.partition_all(2, tuned_opts):
# if okey not in cmd:
# cmd.extend([okey, oval])
# Avoid filtering duplicates on high depth targeted regions where we don't mark duplicates
if any(not dd.get_mark_duplicates(data) for data in items):
cmd += ["--filterDuplicates=0"]
post_process_cmd = (" | %s | %s | %s | vcfallelicprimitives -t DECOMPOSED --keep-geno | vcffixup - | "
"vcfstreamsort | bgzip -c > %s" % (vcfutils.fix_ambiguous_cl(),
vcfutils.fix_ambiguous_cl(5),
vcfutils.add_contig_to_header_cl(items[0]),
tx_out_file))
do.run(" ".join(cmd) + post_process_cmd, "platypus variant calling")
out_file = vcfutils.bgzip_and_index(out_file, items[0]["config"])
return out_file
def merge_bam_files(bam_files, work_dir, data, out_file=None, batch=None):
"""Merge multiple BAM files from a sample into a single BAM for processing.
Checks system open file limit and merges in batches if necessary to avoid
file handle limits.
"""
out_file = _merge_outfile_fname(out_file, bam_files, work_dir, batch)
if not utils.file_exists(out_file):
if len(bam_files) == 1 and bam.bam_already_sorted(bam_files[0], data["config"], "coordinate"):
with file_transaction(data, out_file) as tx_out_file:
_create_merge_filelist(bam_files, tx_out_file, data["config"])
out_file = bam_files[0]
samtools = config_utils.get_program("samtools", data["config"])
do.run('{} quickcheck -v {}'.format(samtools, out_file),
"Check for valid merged BAM after transfer")
else:
with tx_tmpdir(data) as tmpdir:
with utils.chdir(tmpdir):
with file_transaction(data, out_file) as tx_out_file:
tx_bam_file_list = _create_merge_filelist(bam_files, tx_out_file, data["config"])
samtools = config_utils.get_program("samtools", data["config"])
resources = config_utils.get_resources("samtools", data["config"])
num_cores = dd.get_num_cores(data)
# Aim for 3.5Gb/core memory for BAM merging
num_cores = config_utils.adjust_cores_to_mb_target(
3500, resources.get("memory", "2G"), num_cores)
max_mem = config_utils.adjust_memory(resources.get("memory", "1G"),
2, "decrease").upper()
if dd.get_mark_duplicates(data):
cmd = _biobambam_merge_dedup_maxcov(data)
else:
cmd = _biobambam_merge_maxcov(data)
do.run(cmd.format(**locals()), "Merge bam files to %s" % os.path.basename(out_file),
None)
do.run('{} quickcheck -v {}'.format(samtools, tx_out_file),
"Check for valid merged BAM")
do.run('{} quickcheck -v {}'.format(samtools, out_file),
"Check for valid merged BAM after transfer")
_finalize_merge(out_file, bam_files, data["config"])
bam.index(out_file, data["config"])
return out_file
def parallel_prep_region(samples, run_parallel):
"""Perform full pre-variant calling BAM prep work on regions.
"""
file_key = "work_bam"
split_fn = _split_by_regions("bamprep", "-prep.bam", file_key)
# identify samples that do not need preparation -- no recalibration or realignment
extras = []
torun = []
for data in [x[0] for x in samples]:
if data.get("work_bam"):
data["align_bam"] = data["work_bam"]
if (not dd.get_realign(data) and not dd.get_variantcaller(data)):
extras.append([data])
elif not data.get(file_key):
extras.append([data])
else:
# Do not want to re-run duplicate marking after realignment
data["config"]["algorithm"]["orig_markduplicates"] = dd.get_mark_duplicates(data)
data = dd.set_mark_duplicates(data, False)
torun.append([data])
return extras + parallel_split_combine(torun, split_fn, run_parallel,
"piped_bamprep", _add_combine_info, file_key, ["config"])
def parallel_prep_region(samples, run_parallel):
"""Perform full pre-variant calling BAM prep work on regions.
"""
file_key = "work_bam"
split_fn = _split_by_regions("bamprep", "-prep.bam", file_key)
# identify samples that do not need preparation -- no recalibration or realignment
extras = []
torun = []
for data in [x[0] for x in samples]:
if data.get("work_bam"):
data["align_bam"] = data["work_bam"]
if (not dd.get_realign(data) and not dd.get_variantcaller(data)):
extras.append([data])
elif not data.get(file_key):
extras.append([data])
else:
# Do not want to re-run duplicate marking after realignment
data["config"]["algorithm"][
"orig_markduplicates"] = dd.get_mark_duplicates(data)
data = dd.set_mark_duplicates(data, False)
torun.append([data])
return extras + parallel_split_combine(torun, split_fn, run_parallel,
"piped_bamprep", _add_combine_info,
file_key, ["config"])
def _skip_duplicates(data):
return (dd.get_coverage_interval(data) == "amplicon"
or (dd.get_aligner(data) and not dd.get_mark_duplicates(data)))
def _skip_duplicates(data):
return dd.get_coverage_interval(data) == "amplicon" or not dd.get_mark_duplicates(data)
def _skip_duplicates(data):
return (dd.get_coverage_interval(data) == "amplicon" or
(dd.get_aligner(data) and not dd.get_mark_duplicates(data)))
def _skip_duplicates(data):
return dd.get_coverage_interval(
data) == "amplicon" or not dd.get_mark_duplicates(data)