import logging
from dae.gene_sets.denovo_gene_set_collection import DenovoGeneSetCollection
from dae.studies.study import GenotypeData
logger = logging.getLogger(__name__)
[docs]
class DenovoGeneSetHelpers:
"""Helper functions for creation of denovo gene sets."""
[docs]
@classmethod
def load_collection(
cls,
study: GenotypeData,
) -> DenovoGeneSetCollection | None:
"""Load a denovo gene set collection for a given study."""
dgsc = DenovoGeneSetCollection.create_empty_collection(study)
if dgsc is None:
logger.info(
"No denovo gene set collection for %s", study.study_id)
return None
cache_dir = study.config.get("conf_dir")
assert cache_dir is not None
dgsc.load(cache_dir)
return dgsc
[docs]
@classmethod
def load_collection_from_dict(
cls,
study: GenotypeData,
cache: dict,
) -> DenovoGeneSetCollection | None:
"""Load a denovo gene set collection for a given study."""
dgsc = DenovoGeneSetCollection.create_empty_collection(study)
if dgsc is None:
logger.info(
"No denovo gene set collection for %s", study.study_id)
return None
dgsc.cache = dgsc._convert_cache_innermost_types( # noqa: SLF001
cache, list, set,
)
return dgsc
[docs]
@classmethod
def build_collection(
cls, study: GenotypeData, *,
force: bool = False,
) -> DenovoGeneSetCollection | None:
"""Build a denovo gene set collection for a study and save it."""
dgsc = DenovoGeneSetCollection.create_empty_collection(study)
if dgsc is None:
logger.info(
"no denovo gene set collection defined for %s", study.study_id)
return None
cache_dir = study.config.get("conf_dir")
assert cache_dir is not None
if dgsc.is_cached(cache_dir) and not force:
logger.info(
"denovo gene set collection for %s already cached",
study.study_id,
)
dgsc.load(cache_dir)
return dgsc
dgsc = DenovoGeneSetCollection.build_collection(study)
if dgsc is None:
return None
dgsc.save(cache_dir)
return dgsc