def main(argv=sys.argv[1:], gpf_instance=None):
if gpf_instance is None:
gpf_instance = GPFInstance()
argv = parse_cli_arguments(argv, gpf_instance)
genotype_storage_db = gpf_instance.genotype_storage_db
genotype_storage = genotype_storage_db.get_genotype_storage(
argv.genotype_storage
)
if not genotype_storage or (
genotype_storage and not genotype_storage.is_impala()):
print("missing or non-impala genotype storage")
return
assert os.path.exists(argv.variants)
study_config = genotype_storage.impala_load_dataset(
argv.study_id, argv.variants, argv.pedigree)
if argv.study_config:
input_config = GPFConfigParser.load_config_raw(argv.study_config)
study_config = recursive_dict_update(study_config, input_config)
study_config = StudyConfigBuilder(study_config).build_config()
assert study_config is not None
save_study_config(
gpf_instance.dae_config, argv.study_id, study_config,
force=argv.force)
def simple_study_import(
self,
study_id,
families_loader=None,
variant_loaders=None,
study_config=None,
**kwargs,
):
families_config = self._import_families_file(study_id, families_loader)
variants_config = self._import_variants_files(study_id,
variant_loaders)
config_dict = {
"id": study_id,
"conf_dir": ".",
"has_denovo": False,
"has_cnv": False,
"genotype_storage": {
"id": self.id,
"files": {
"variants": variants_config,
"pedigree": families_config,
},
},
"genotype_browser": {
"enabled": True
},
}
if not variant_loaders:
config_dict["genotype_browser"]["enabled"] = False
else:
variant_loaders[0].get_attribute("source_type")
if any([
loader.get_attribute("source_type") == "denovo"
for loader in variant_loaders
]):
config_dict["has_denovo"] = True
if any([
loader.get_attribute("source_type") == "cnv"
for loader in variant_loaders
]):
config_dict["has_denovo"] = True
config_dict["has_cnv"] = True
if study_config is not None:
study_config_dict = GPFConfigParser.load_config_raw(study_config)
config_dict = recursive_dict_update(config_dict, study_config_dict)
config_builder = StudyConfigBuilder(config_dict)
return config_builder.build_config()
def simple_study_import(self,
study_id,
families_loader=None,
variant_loaders=None,
study_config=None,
output=".",
include_reference=False):
variants_dir = None
has_denovo = False
has_cnv = False
bucket_index = 0
if variant_loaders:
for index, variant_loader in enumerate(variant_loaders):
assert isinstance(variant_loader, VariantsLoader), \
type(variant_loader)
if variant_loader.get_attribute("source_type") == "denovo":
has_denovo = True
if variant_loader.get_attribute("source_type") == "cnv":
has_denovo = True
has_cnv = True
if variant_loader.transmission_type == \
TransmissionType.denovo:
assert index < 100
bucket_index = index # denovo buckets < 100
elif variant_loader.transmission_type == \
TransmissionType.transmitted:
bucket_index = index + 100 # transmitted buckets >=100
variants_dir = os.path.join(output, "variants")
partition_description = NoPartitionDescriptor(variants_dir)
ParquetManager.variants_to_parquet(
variant_loader,
partition_description,
# parquet_filenames.variants,
bucket_index=bucket_index,
include_reference=include_reference)
pedigree_filename = os.path.join(output, "pedigree",
"pedigree.parquet")
families = families_loader.load()
ParquetManager.families_to_parquet(families, pedigree_filename)
config_dict = self.impala_load_dataset(study_id,
variants_dir=variants_dir,
pedigree_file=pedigree_filename)
config_dict["has_denovo"] = has_denovo
config_dict["has_cnv"] = has_cnv
if study_config is not None:
study_config_dict = GPFConfigParser.load_config_raw(study_config)
config_dict = recursive_dict_update(config_dict, study_config_dict)
config_builder = StudyConfigBuilder(config_dict)
return config_builder.build_config()