def test_load_family_simple(fixture_name, temp_filename, fixture_dirname):
family_filename = fixture_dirname(fixture_name)
assert os.path.exists(family_filename)
families = FamiliesLoader.load_simple_families_file(family_filename)
assert families is not None
FamiliesLoader.save_pedigree(families, temp_filename)
families1 = FamiliesLoader.load_pedigree_file(temp_filename)
assert set(families.keys()) == set(families1.keys())
def main(argv):
args = parse_cli_arguments(argv[1:])
if args.id is not None:
study_id = args.id
else:
study_id, _ = os.path.splitext(os.path.basename(args.family_filename))
if args.output is None:
output = "{study_id}.ped".format(study_id=study_id)
else:
output = args.output
fam_df = FamiliesLoader.load_simple_family_file(args.family_filename)
FamiliesLoader.save_pedigree(fam_df, output)
def main(argv, gpf_instance=None):
if gpf_instance is None:
gpf_instance = GPFInstance()
parser = argparse.ArgumentParser()
parser.add_argument('--verbose', '-V', action='count', default=0)
FamiliesLoader.cli_arguments(parser)
VcfLoader.cli_arguments(parser, options_only=True)
parser.add_argument(
"-o",
"--output",
dest="output_filename",
help="output families parquet filename "
"(default is [basename(families_filename).ped])",
)
parser.add_argument(
"--partition-description",
"--pd",
help="input partition description filename",
)
parser.add_argument(
"--vcf-files",
type=str,
nargs="+",
metavar="<VCF filename>",
help="VCF file to import",
)
argv = parser.parse_args(argv)
if argv.verbose == 1:
logging.basicConfig(level=logging.WARNING)
elif argv.verbose == 2:
logging.basicConfig(level=logging.INFO)
elif argv.verbose >= 3:
logging.basicConfig(level=logging.DEBUG)
else:
logging.basicConfig(level=logging.WARNING)
filename, params = FamiliesLoader.parse_cli_arguments(argv)
logger.info(F"PED PARAMS: {params}")
loader = FamiliesLoader(filename, **params)
families = loader.load()
if argv.partition_description:
partition_description = ParquetPartitionDescriptor.from_config(
argv.partition_description)
families = partition_description.add_family_bins_to_families(families)
variants_filenames, variants_params = \
VcfLoader.parse_cli_arguments(argv)
if variants_filenames:
assert variants_filenames is not None
variants_loader = VcfLoader(
families,
variants_filenames,
params=variants_params,
genome=gpf_instance.genomes_db.get_genome(),
)
families = variants_loader.families
if families.broken_families:
for family_id, family in families.broken_families.items():
if not family.has_members():
del families[family_id]
logger.warning(
f"family {family_id} does not contain sequenced members "
f"and is removed from the pedigree: {family}")
if not argv.output_filename:
output_filename, _ = os.path.splitext(os.path.basename(filename))
output_filename = f"{output_filename}.ped"
else:
output_filename = argv.output_filename
FamiliesLoader.save_pedigree(families, output_filename)