sfaira.data.load_store

sfaira.data.load_store(cache_path: Union[str, os.PathLike, List[str], List[os.PathLike]], store_format: str = 'dao', columns: Optional[List[str]] = None) → sfaira.data.store.stores.multi.StoreMultipleFeatureSpaceBase

Instantiates a distributed store class.

Note that any store is instantiated as a DistributedStoreMultipleFeatureSpaceBase. This instances can be subsetted to the desired single feature space.

Parameters

• cache_path – Store directory.

• store_format –

  Format of store {“h5ad”, “dao”}.

  • ”h5ad”: Returns instance of DistributedStoreH5ad and keeps data in memory. See also “h5ad_backed”.

  • ”dao”: Returns instance of DistributedStoreDoa (distributed access optimized).

  • ”h5ad_backed”: Returns instance of DistributedStoreH5ad and keeps data as backed (out of memory). See also

    ”h5ad”.

• columns – Which columns to read into the obs copy in the output, see pandas.read_parquet(). Only relevant if store_format is “dao”.

Returns

Instances of a distributed store class.