def _run_toplevel(config,
config_file,
work_dir,
parallel,
fc_dir=None,
run_info_yaml=None):
"""
Run toplevel analysis, processing a set of input files.
config_file -- Main YAML configuration file with system parameters
fc_dir -- Directory of fastq files to process
run_info_yaml -- YAML configuration file specifying inputs to process
"""
parallel = log.create_base_logger(config, parallel)
log.setup_local_logging(config, parallel)
logger.info("System YAML configuration: %s" % os.path.abspath(config_file))
dirs = run_info.setup_directories(work_dir, fc_dir, config, config_file)
config_file = os.path.join(dirs["config"], os.path.basename(config_file))
pipelines, config = _pair_samples_with_pipelines(run_info_yaml, config)
system.write_info(dirs, parallel, config)
with tx_tmpdir(config if parallel.get("type") ==
"local" else None) as tmpdir:
tempfile.tempdir = tmpdir
for pipeline, samples in pipelines.items():
for xs in pipeline(config, run_info_yaml, parallel, dirs, samples):
pass
def _run_toplevel(config,
config_file,
work_dir,
parallel,
fc_dir=None,
run_info_yaml=None):
"""
Run toplevel analysis, processing a set of input files.
config_file -- Main YAML configuration file with system parameters
fc_dir -- Directory of fastq files to process
run_info_yaml -- YAML configuration file specifying inputs to process
"""
parallel = log.create_base_logger(config, parallel)
log.setup_local_logging(config, parallel)
dirs = run_info.setup_directories(work_dir, fc_dir, config, config_file)
config_file = os.path.join(dirs["config"], os.path.basename(config_file))
pipelines = _pair_samples_with_pipelines(run_info_yaml)
system.write_info(dirs, parallel, config)
final = []
with tx_tmpdir(config) as tmpdir:
tempfile.tempdir = tmpdir
for pipeline, samples in pipelines.items():
for xs in pipeline.run(config, run_info_yaml, parallel, dirs,
samples):
if len(xs) == 1:
upload.from_sample(xs[0])
final.append(xs[0])
def _run_toplevel(config, config_file, work_dir, parallel,
fc_dir=None, run_info_yaml=None, samples=None):
"""
Run toplevel analysis, processing a set of input files.
config_file -- Main YAML configuration file with system parameters
fc_dir -- Directory of fastq files to process
run_info_yaml -- YAML configuration file specifying inputs to process
samples -- Pre-processed samples, useful if run inside of docker containers.
"""
parallel = log.create_base_logger(config, parallel)
log.setup_local_logging(config, parallel)
dirs = run_info.setup_directories(work_dir, fc_dir, config, config_file)
config_file = os.path.join(dirs["config"], os.path.basename(config_file))
if samples:
dockerized = True
else:
dockerized = False
samples = run_info.organize(dirs, config, run_info_yaml)
pipelines = _pair_samples_with_pipelines(samples)
final = []
with tx_tmpdir(config) as tmpdir:
tempfile.tempdir = tmpdir
for pipeline, pipeline_items in pipelines.items():
pipeline_items = _add_provenance(pipeline_items, dirs, parallel, config)
if not dockerized:
versioncheck.testall(pipeline_items)
for xs in pipeline.run(config, config_file, parallel, dirs, pipeline_items):
if len(xs) == 1:
upload.from_sample(xs[0])
final.append(xs[0])
def create_sample_config(data_dir, work_dir, disambiguate=False):
system_config, system_file = config_utils.load_system_config(work_dir=work_dir)
system_config["dirs"] = run_info.setup_directories(work_dir, work_dir, system_config, system_file)
c = ConfigCreator(data_dir, work_dir, system_config)
if disambiguate:
return c.config_with_disambiguate()
else:
return c.config_without_disambiguate()
def _run_toplevel(config, config_file, work_dir, parallel, fc_dir=None, run_info_yaml=None):
"""
Run toplevel analysis, processing a set of input files.
config_file -- Main YAML configuration file with system parameters
fc_dir -- Directory of fastq files to process
run_info_yaml -- YAML configuration file specifying inputs to process
"""
parallel = log.create_base_logger(config, parallel)
log.setup_local_logging(config, parallel)
dirs = run_info.setup_directories(work_dir, fc_dir, config, config_file)
config_file = os.path.join(dirs["config"], os.path.basename(config_file))
pipelines, config = _pair_samples_with_pipelines(run_info_yaml, config)
system.write_info(dirs, parallel, config)
with tx_tmpdir(config) as tmpdir:
tempfile.tempdir = tmpdir
for pipeline, samples in pipelines.items():
for xs in pipeline.run(config, run_info_yaml, parallel, dirs, samples):
pass
