def _run_delly(bam_files, chrom, ref_file, work_dir, items):
"""Run delly, calling structural variations for the specified type.
"""
batch = sshared.get_cur_batch(items)
ext = "-%s-svs" % batch if batch else "-svs"
out_file = os.path.join(
work_dir, "%s%s-%s.bcf" %
(os.path.splitext(os.path.basename(bam_files[0]))[0], ext, chrom))
final_file = "%s.vcf.gz" % (utils.splitext_plus(out_file)[0])
cores = min(
utils.get_in(items[0], ("config", "algorithm", "num_cores"), 1),
len(bam_files))
if not utils.file_exists(out_file) and not utils.file_exists(final_file):
with file_transaction(items[0], out_file) as tx_out_file:
if sshared.has_variant_regions(items, out_file, chrom):
exclude = ["-x", _delly_exclude_file(items, out_file, chrom)]
cmd = ["delly", "call", "-g", ref_file, "-o", tx_out_file
] + exclude + bam_files
locale_to_use = utils.get_locale()
multi_cmd = "export OMP_NUM_THREADS=%s && export LC_ALL=%s && " % (
cores, locale_to_use)
try:
do.run(multi_cmd + " ".join(cmd),
"delly structural variant")
except subprocess.CalledProcessError as msg:
# Small input samples, write an empty vcf
if "Sample has not enough data to estimate library parameters" in str(
msg):
pass
# delly returns an error exit code if there are no variants
elif "No structural variants found" not in str(msg):
raise
return [_bgzip_and_clean(out_file, items)]
def _setup_logging(args):
# Set environment to standard to use periods for decimals and avoid localization
locale_to_use = utils.get_locale()
os.environ["LC_ALL"] = locale_to_use
os.environ["LC"] = locale_to_use
os.environ["LANG"] = locale_to_use
config = None
if len(args) == 1 and isinstance(args[0], (list, tuple)):
args = args[0]
for arg in args:
if config_utils.is_nested_config_arg(arg):
config = arg["config"]
break
elif config_utils.is_std_config_arg(arg):
config = arg
break
elif isinstance(arg,
(list, tuple)) and config_utils.is_nested_config_arg(
arg[0]):
config = arg[0]["config"]
break
if config is None:
raise NotImplementedError("No config found in arguments: %s" % args[0])
handler = setup_local_logging(config, config.get("parallel", {}))
try:
yield config
except:
logger.exception("Unexpected error")
raise
finally:
if hasattr(handler, "close"):
handler.close()
def run_main(workdir, config_file=None, fc_dir=None, run_info_yaml=None,
parallel=None, workflow=None):
"""Run variant analysis, handling command line options.
"""
# Set environment to standard to use periods for decimals and avoid localization
locale_to_use = utils.get_locale()
os.environ["LC_ALL"] = locale_to_use
os.environ["LC"] = locale_to_use
os.environ["LANG"] = locale_to_use
workdir = utils.safe_makedir(os.path.abspath(workdir))
os.chdir(workdir)
config, config_file = config_utils.load_system_config(config_file, workdir)
parallel = log.create_base_logger(config, parallel)
log.setup_local_logging(config, parallel)
logger.info(f"System YAML configuration: {os.path.abspath(config_file)}.")
logger.info(f"Locale set to {locale_to_use}.")
if config.get("log_dir", None) is None:
config["log_dir"] = os.path.join(workdir, DEFAULT_LOG_DIR)
if parallel["type"] in ["local", "clusterk"]:
_setup_resources()
_run_toplevel(config, config_file, workdir, parallel,
fc_dir, run_info_yaml)
elif parallel["type"] == "ipython":
assert parallel["scheduler"] is not None, "IPython parallel requires a specified scheduler (-s)"
if parallel["scheduler"] != "sge":
assert parallel["queue"] is not None, "IPython parallel requires a specified queue (-q)"
elif not parallel["queue"]:
parallel["queue"] = ""
_run_toplevel(config, config_file, workdir, parallel,
fc_dir, run_info_yaml)
else:
raise ValueError("Unexpected type of parallel run: %s" % parallel["type"])
def process(args):
"""Run the function in args.name given arguments in args.argfile.
"""
# Set environment to standard to use periods for decimals and avoid localization
locale_to_use = utils.get_locale()
os.environ["LC_ALL"] = locale_to_use
os.environ["LC"] = locale_to_use
os.environ["LANG"] = locale_to_use
setpath.prepend_bcbiopath()
try:
fn = getattr(multitasks, args.name)
except AttributeError:
raise AttributeError(
"Did not find exposed function in bcbio.distributed.multitasks named '%s'"
% args.name)
if args.moreargs or args.raw:
fnargs = [args.argfile] + args.moreargs
work_dir = None
argfile = None
else:
with open(args.argfile) as in_handle:
fnargs = yaml.safe_load(in_handle)
work_dir = os.path.dirname(args.argfile)
fnargs = config_utils.merge_resources(fnargs)
argfile = args.outfile if args.outfile else "%s-out%s" % os.path.splitext(
args.argfile)
if not work_dir:
work_dir = os.getcwd()
if len(fnargs) > 0 and fnargs[0] == "cwl":
fnargs, parallel, out_keys, input_files = _world_from_cwl(
args.name, fnargs[1:], work_dir)
# Can remove this awkward Docker merge when we do not need custom GATK3 installs
fnargs = config_utils.merge_resources(fnargs)
argfile = os.path.join(work_dir, "cwl.output.json")
else:
parallel, out_keys, input_files = None, {}, []
with utils.chdir(work_dir):
with contextlib.closing(
log.setup_local_logging(parallel={"wrapper": "runfn"})):
try:
out = fn(*fnargs)
except:
logger.exception()
raise
finally:
# Clean up any copied and unpacked workflow inputs, avoiding extra disk usage
wf_input_dir = os.path.join(work_dir, "wf-inputs")
if os.path.exists(wf_input_dir) and os.path.isdir(
wf_input_dir):
shutil.rmtree(wf_input_dir)
if argfile:
try:
_write_out_argfile(argfile, out, fnargs, parallel, out_keys,
input_files, work_dir)
except:
logger.exception()
raise