def _sambamba_merge(bam_files):
"""Merge multiple BAM files with sambamba.
"""
if len(bam_files) > system.open_file_limit():
raise IOError("More files to merge (%s) than available open file descriptors (%s)\n"
"See documentation on tips for changing file limits:\n"
"https://bcbio-nextgen.readthedocs.org/en/latest/contents/"
"parallel.html#tuning-systems-for-scale"
% (len(bam_files), system.open_file_limit()))
return "{sambamba} merge {tx_out_file} -t {num_cores} `cat {tx_bam_file_list}`"
def _sambamba_merge(bam_files):
"""Merge multiple BAM files with sambamba.
"""
if len(bam_files) > system.open_file_limit():
raise IOError("More files to merge (%s) than available open file descriptors (%s)\n"
"See documentation on tips for changing file limits:\n"
"https://bcbio-nextgen.readthedocs.org/en/latest/contents/"
"parallel.html#tuning-systems-for-scale"
% (len(bam_files), system.open_file_limit()))
return "{sambamba} merge {tx_out_file} -t {num_cores} `cat {tx_bam_file_list}`"
def _bamtools_merge(bam_files):
"""Use bamtools to merge multiple BAM files, requires a list from disk.
"""
if len(bam_files) > system.open_file_limit():
raise IOError("More files to merge (%s) then available open file descriptors (%s)\n"
"See documentation on tips for changing file limits:\n"
"https://bcbio-nextgen.readthedocs.org/en/latest/contents/"
"parallel.html#tuning-systems-for-scale"
% (len(bam_files), system.open_file_limit()))
return "{bamtools} merge -list {bam_file_list}"
def _bamtools_merge(bam_files):
"""Use bamtools to merge multiple BAM files, requires a list from disk.
"""
if len(bam_files) > system.open_file_limit():
raise IOError("More files to merge (%s) than available open file descriptors (%s)\n"
"See documentation on tips for changing file limits:\n"
"https://bcbio-nextgen.readthedocs.org/en/latest/contents/"
"parallel.html#tuning-systems-for-scale"
% (len(bam_files), system.open_file_limit()))
return "{bamtools} merge -list {bam_file_list}"
def _samtools_merge(bam_files):
"""Concatenate multiple BAM files together with samtools.
Creates short paths to shorten the commandline.
"""
if len(bam_files) > system.open_file_limit():
raise IOError("More files to merge (%s) than available open file descriptors (%s)\n"
"See documentation on tips for changing file limits:\n"
"https://bcbio-nextgen.readthedocs.org/en/latest/contents/"
"parallel.html#tuning-systems-for-scale"
% (len(bam_files), system.open_file_limit()))
return "{samtools} merge - `cat {tx_bam_file_list}`"
def _samtools_merge(bam_files):
"""Concatenate multiple BAM files together with samtools.
Creates short paths to shorten the commandline.
"""
if len(bam_files) > system.open_file_limit():
raise IOError(
"More files to merge (%s) than available open file descriptors (%s)\n"
"See documentation on tips for changing file limits:\n"
"https://bcbio-nextgen.readthedocs.org/en/latest/contents/"
"parallel.html#tuning-systems-for-scale" %
(len(bam_files), system.open_file_limit()))
return "{samtools} merge - `cat {tx_bam_file_list}`"
def merge_bam_files(bam_files, work_dir, config, 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.
"""
if len(bam_files) == 1 and bam.bam_already_sorted(bam_files[0], config,
"coordinate"):
shutil.copy(bam_files[0], out_file)
else:
if out_file is None:
out_file = os.path.join(work_dir,
os.path.basename(sorted(bam_files)[0]))
if batch is not None:
base, ext = os.path.splitext(out_file)
out_file = "%s-b%s%s" % (base, batch, ext)
if not utils.file_exists(out_file):
sambamba = config_utils.get_program("sambamba", config)
samtools = config_utils.get_program("samtools", config)
resources = config_utils.get_resources("samtools", config)
num_cores = config["algorithm"].get("num_cores", 1)
max_mem = config_utils.adjust_memory(resources.get("memory", "1G"),
2, "decrease").upper()
# sambamba opens 4 handles per file, so try to guess a reasonable batch size
batch_size = (system.open_file_limit() // 4) - 100
if len(bam_files) > batch_size:
bam_files = [
merge_bam_files(xs, work_dir, config, out_file, i)
for i, xs in enumerate(
utils.partition_all(batch_size, bam_files))
]
with tx_tmpdir(config) as tmpdir:
with utils.chdir(tmpdir):
with file_transaction(config, out_file) as tx_out_file:
with file_transaction(
config,
"%s.list" % os.path.splitext(out_file)[0]
) as tx_bam_file_list:
with open(tx_bam_file_list, "w") as out_handle:
for f in sorted(bam_files):
out_handle.write("%s\n" % f)
if bam.bam_already_sorted(bam_files[0], config,
"coordinate"):
cmd = _sambamba_merge(bam_files)
else:
assert config.get("mark_duplicates", True)
cmd = _biobambam_merge_dedup()
do.run(
cmd.format(**locals()),
"Merge bam files to %s" %
os.path.basename(out_file), None)
# Ensure timestamps are up to date on output file and index
# Works around issues on systems with inconsistent times
for ext in ["", ".bai"]:
if os.path.exists(out_file + ext):
subprocess.check_call(["touch", out_file + ext])
for b in bam_files:
utils.save_diskspace(b, "BAM merged to %s" % out_file, config)
bam.index(out_file, config)
return out_file
def merge_bam_files(bam_files, work_dir, config, 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], config, "coordinate"):
with file_transaction(config, out_file) as tx_out_file:
_create_merge_filelist(bam_files, tx_out_file, config)
shutil.copy(bam_files[0], tx_out_file)
samtools = config_utils.get_program("samtools", config)
do.run('{} quickcheck -v {}'.format(samtools, out_file),
"Check for valid merged BAM after transfer")
else:
# sambamba opens 4 handles per file, so try to guess a reasonable batch size
batch_size = (system.open_file_limit() // 4) - 100
if len(bam_files) > batch_size:
bam_files = [
merge_bam_files(xs, work_dir, config, out_file, i)
for i, xs in enumerate(
utils.partition_all(batch_size, bam_files))
]
with tx_tmpdir(config) as tmpdir:
with utils.chdir(tmpdir):
with file_transaction(config, out_file) as tx_out_file:
tx_bam_file_list = _create_merge_filelist(
bam_files, tx_out_file, config)
sambamba = config_utils.get_program("sambamba", config)
samtools = config_utils.get_program("samtools", config)
resources = config_utils.get_resources(
"samtools", config)
num_cores = config["algorithm"].get("num_cores", 1)
max_mem = config_utils.adjust_memory(
resources.get("memory", "1G"), 2,
"decrease").upper()
if bam.bam_already_sorted(bam_files[0], config,
"coordinate"):
cmd = _sambamba_merge(bam_files)
else:
assert config.get("mark_duplicates", True)
cmd = _biobambam_merge_dedup()
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, config)
bam.index(out_file, config)
return out_file
def merge_bam_files(bam_files, work_dir, config, out_file=None):
"""Merge multiple BAM files from a sample into a single BAM for processing.
Uses bamtools for merging, which handles large numbers of input BAMs.
"""
if len(bam_files) == 1:
return bam_files[0]
else:
if out_file is None:
out_file = os.path.join(work_dir,
os.path.basename(sorted(bam_files)[0]))
if not utils.file_exists(out_file) or not utils.file_exists(out_file +
".bai"):
bamtools = config_utils.get_program("bamtools", config)
samtools = config_utils.get_program("samtools", config)
resources = config_utils.get_resources("samtools", config)
num_cores = config["algorithm"].get("num_cores", 1)
max_mem = resources.get("memory", "1G")
if len(bam_files) > system.open_file_limit():
raise IOError(
"More files to merge (%s) then available open file descriptors (%s)\n"
"See documentation on tips for changing file limits:\n"
"https://bcbio-nextgen.readthedocs.org/en/latest/contents/"
"parallel.html#tuning-systems-for-scale" %
(len(bam_files), system.open_file_limit()))
with file_transaction(out_file) as tx_out_file:
tx_out_prefix = os.path.splitext(tx_out_file)[0]
with utils.tmpfile(dir=work_dir,
prefix="bammergelist") as bam_file_list:
bam_file_list = "%s.list" % os.path.splitext(out_file)[0]
with open(bam_file_list, "w") as out_handle:
for f in sorted(bam_files):
out_handle.write("%s\n" % f)
cmd = (
"{bamtools} merge -list {bam_file_list} | "
"{samtools} sort -@ {num_cores} -m {max_mem} - {tx_out_prefix}"
)
do.run(cmd.format(**locals()), "Merge bam files", None)
for b in bam_files:
utils.save_diskspace(b, "BAM merged to %s" % out_file, config)
picard = broad.runner_from_config(config)
picard.run_fn("picard_index", out_file)
return out_file
def merge_bam_files(bam_files, work_dir, config, 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.
"""
if len(bam_files) == 1:
return bam_files[0]
else:
if out_file is None:
out_file = os.path.join(work_dir,
os.path.basename(sorted(bam_files)[0]))
if batch is not None:
base, ext = os.path.splitext(out_file)
out_file = "%s-b%s%s" % (base, batch, ext)
if not utils.file_exists(out_file) or not utils.file_exists(out_file +
".bai"):
bamtools = config_utils.get_program("bamtools", config)
samtools = config_utils.get_program("samtools", config)
resources = config_utils.get_resources("samtools", config)
num_cores = config["algorithm"].get("num_cores", 1)
max_mem = config_utils.adjust_memory(resources.get("memory", "1G"),
2, "decrease").upper()
batch_size = system.open_file_limit() - 100
if len(bam_files) > batch_size:
bam_files = [
merge_bam_files(xs, work_dir, config, out_file, i)
for i, xs in enumerate(
utils.partition_all(batch_size, bam_files))
]
with utils.curdir_tmpdir({"config": config}) as tmpdir:
with utils.chdir(tmpdir):
merge_cl = _bamtools_merge(bam_files)
with file_transaction(out_file) as tx_out_file:
with file_transaction("%s.list" %
os.path.splitext(out_file)[0]
) as tx_bam_file_list:
tx_out_prefix = os.path.splitext(tx_out_file)[0]
with open(tx_bam_file_list, "w") as out_handle:
for f in sorted(bam_files):
out_handle.write("%s\n" % f)
cmd = (
merge_cl + " | "
"{samtools} sort -@ {num_cores} -m {max_mem} - {tx_out_prefix}"
)
do.run(
cmd.format(**locals()),
"Merge bam files to %s" %
os.path.basename(out_file), None)
for b in bam_files:
utils.save_diskspace(b, "BAM merged to %s" % out_file, config)
bam.index(out_file, config)
return out_file
def merge_bam_files(bam_files, work_dir, config, 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.
"""
if len(bam_files) == 1:
bam.index(bam_files[0], config)
return bam_files[0]
else:
if out_file is None:
out_file = os.path.join(work_dir, os.path.basename(sorted(bam_files)[0]))
if batch is not None:
base, ext = os.path.splitext(out_file)
out_file = "%s-b%s%s" % (base, batch, ext)
if not utils.file_exists(out_file):
sambamba = config_utils.get_program("sambamba", config)
samtools = config_utils.get_program("samtools", config)
samblaster = config_utils.get_program("samblaster", config)
resources = config_utils.get_resources("samtools", config)
num_cores = config["algorithm"].get("num_cores", 1)
max_mem = config_utils.adjust_memory(resources.get("memory", "1G"),
2, "decrease").upper()
# sambamba opens 4 handles per file, so try to guess a reasonable batch size
batch_size = (system.open_file_limit() // 4) - 100
if len(bam_files) > batch_size:
bam_files = [merge_bam_files(xs, work_dir, config, out_file, i)
for i, xs in enumerate(utils.partition_all(batch_size, bam_files))]
with tx_tmpdir(config) as tmpdir:
with utils.chdir(tmpdir):
with file_transaction(config, out_file) as tx_out_file:
with file_transaction(config, "%s.list" % os.path.splitext(out_file)[0]) as tx_bam_file_list:
with open(tx_bam_file_list, "w") as out_handle:
for f in sorted(bam_files):
out_handle.write("%s\n" % f)
if bam.bam_already_sorted(bam_files[0], config, "coordinate"):
cmd = _sambamba_merge(bam_files)
else:
assert config.get("mark_duplicates", True)
cmd = _biobambam_merge_dedup()
do.run(cmd.format(**locals()), "Merge bam files to %s" % os.path.basename(out_file),
None)
# Ensure timestamps are up to date on output file and index
# Works around issues on systems with inconsistent times
for ext in ["", ".bai"]:
if os.path.exists(out_file + ext):
subprocess.check_call(["touch", out_file + ext])
for b in bam_files:
utils.save_diskspace(b, "BAM merged to %s" % out_file, config)
bam.index(out_file, config)
return out_file
def merge_bam_files(bam_files, work_dir, config, out_file=None):
"""Merge multiple BAM files from a sample into a single BAM for processing.
Uses bamtools for merging, which handles large numbers of input BAMs.
"""
if len(bam_files) == 1:
return bam_files[0]
else:
if out_file is None:
out_file = os.path.join(work_dir, os.path.basename(sorted(bam_files)[0]))
if not utils.file_exists(out_file) or not utils.file_exists(out_file + ".bai"):
bamtools = config_utils.get_program("bamtools", config)
samtools = config_utils.get_program("samtools", config)
resources = config_utils.get_resources("samtools", config)
num_cores = config["algorithm"].get("num_cores", 1)
max_mem = resources.get("memory", "1G")
if len(bam_files) > system.open_file_limit():
raise IOError("More files to merge (%s) then available open file descriptors (%s)\n"
"See documentation on tips for changing file limits:\n"
"https://bcbio-nextgen.readthedocs.org/en/latest/contents/"
"parallel.html#tuning-systems-for-scale"
% (len(bam_files), system.open_file_limit()))
with file_transaction(out_file) as tx_out_file:
tx_out_prefix = os.path.splitext(tx_out_file)[0]
with utils.tmpfile(dir=work_dir, prefix="bammergelist") as bam_file_list:
bam_file_list = "%s.list" % os.path.splitext(out_file)[0]
with open(bam_file_list, "w") as out_handle:
for f in sorted(bam_files):
out_handle.write("%s\n" % f)
cmd = ("{bamtools} merge -list {bam_file_list} | "
"{samtools} sort -@ {num_cores} -m {max_mem} - {tx_out_prefix}")
do.run(cmd.format(**locals()), "Merge bam files", None)
for b in bam_files:
utils.save_diskspace(b, "BAM merged to %s" % out_file, config)
picard = broad.runner_from_config(config)
picard.run_fn("picard_index", out_file)
return out_file
def merge_bam_files(bam_files, work_dir, config, 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], config, "coordinate"):
with file_transaction(config, out_file) as tx_out_file:
_create_merge_filelist(bam_files, tx_out_file, config)
shutil.copy(bam_files[0], tx_out_file)
samtools = config_utils.get_program("samtools", config)
do.run('{} quickcheck -v {}'.format(samtools, out_file),
"Check for valid merged BAM after transfer")
else:
# sambamba opens 4 handles per file, so try to guess a reasonable batch size
batch_size = (system.open_file_limit() // 4) - 100
if len(bam_files) > batch_size:
bam_files = [merge_bam_files(xs, work_dir, config, out_file, i)
for i, xs in enumerate(utils.partition_all(batch_size, bam_files))]
with tx_tmpdir(config) as tmpdir:
with utils.chdir(tmpdir):
with file_transaction(config, out_file) as tx_out_file:
tx_bam_file_list = _create_merge_filelist(bam_files, tx_out_file, config)
sambamba = config_utils.get_program("sambamba", config)
samtools = config_utils.get_program("samtools", config)
resources = config_utils.get_resources("samtools", config)
num_cores = config["algorithm"].get("num_cores", 1)
max_mem = config_utils.adjust_memory(resources.get("memory", "1G"),
2, "decrease").upper()
if bam.bam_already_sorted(bam_files[0], config, "coordinate"):
cmd = _sambamba_merge(bam_files)
else:
# 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)
assert config.get("mark_duplicates", True)
cmd = _biobambam_merge_dedup()
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, config)
bam.index(out_file, config)
return out_file
def merge_bam_files(bam_files, work_dir, config, 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.
"""
if len(bam_files) == 1:
return bam_files[0]
else:
if out_file is None:
out_file = os.path.join(work_dir, os.path.basename(sorted(bam_files)[0]))
if batch is not None:
base, ext = os.path.splitext(out_file)
out_file = "%s-b%s%s" % (base, batch, ext)
if not utils.file_exists(out_file) or not utils.file_exists(out_file + ".bai"):
bamtools = config_utils.get_program("bamtools", config)
samtools = config_utils.get_program("samtools", config)
resources = config_utils.get_resources("samtools", config)
num_cores = config["algorithm"].get("num_cores", 1)
max_mem = config_utils.adjust_memory(resources.get("memory", "1G"),
2, "decrease")
batch_size = system.open_file_limit() - 100
if len(bam_files) > batch_size:
bam_files = [merge_bam_files(xs, work_dir, config, out_file, i)
for i, xs in enumerate(utils.partition_all(batch_size, bam_files))]
with utils.curdir_tmpdir({"config": config}) as tmpdir:
with utils.chdir(tmpdir):
merge_cl = _bamtools_merge(bam_files)
with file_transaction(out_file) as tx_out_file:
with file_transaction("%s.list" % os.path.splitext(out_file)[0]) as tx_bam_file_list:
tx_out_prefix = os.path.splitext(tx_out_file)[0]
with open(tx_bam_file_list, "w") as out_handle:
for f in sorted(bam_files):
out_handle.write("%s\n" % f)
cmd = (merge_cl + " | "
"{samtools} sort -@ {num_cores} -m {max_mem} - {tx_out_prefix}")
do.run(cmd.format(**locals()), "Merge bam files to %s" % os.path.basename(out_file),
None)
for b in bam_files:
utils.save_diskspace(b, "BAM merged to %s" % out_file, config)
bam.index(out_file, config)
return out_file
