moscot.utils.tagged_array.DistributionContainer.from_adata

classmethod DistributionContainer.from_adata(adata, a, b, xy_attr=None, xy_key=None, xy_cost='sq_euclidean', xx_attr=None, xx_key=None, xx_cost='sq_euclidean', conditions_attr=None, conditions_key=None, backend='ott', **kwargs)[source]

Create distribution container from AnnData.

Warning

Sparse arrays will be always densified.

Parameters:

  • adata (AnnData) – Annotated data object.

  • a (Union[ndarray[tuple[int, ...], dtype[floating]], Array]) – Marginals when used as source distribution.

  • b (Union[ndarray[tuple[int, ...], dtype[floating]], Array]) – Marginals when used as target distribution.

  • xy_attr (Optional[Literal['X', 'obsp', 'obsm', 'layers', 'uns']]) – Attribute of adata containing the data for the shared space.

  • xy_key (Optional[str]) – Key of xy_attr containing the data for the shared space.

  • xy_cost (Union[str, Literal['barcode_distance', 'leaf_distance', 'custom'], Literal['euclidean', 'sq_euclidean', 'cosine', 'pnorm_p', 'sq_pnorm', 'geodesic']]) – Cost function when in the shared space.

  • xx_attr (Optional[Literal['X', 'obsp', 'obsm', 'layers', 'uns']]) – Attribute of adata containing the data for the incomparable space.

  • xx_key (Optional[str]) – Key of xx_attr containing the data for the incomparable space.

  • xx_cost (Union[str, Literal['barcode_distance', 'leaf_distance', 'custom'], Literal['euclidean', 'sq_euclidean', 'cosine', 'pnorm_p', 'sq_pnorm', 'geodesic']]) – Cost function in the incomparable space.

  • conditions_attr (Optional[Literal['obs', 'var', 'obsm', 'varm', 'layers', 'uns']]) – Attribute of adata containing the conditions.

  • conditions_key (Optional[str]) – Key of conditions_attr containing the conditions.

  • backend (Literal['ott']) – Backend to use.

  • kwargs (Any) – Keyword arguments to pass to the cost functions.

Return type:

DistributionContainer

Returns:

: The distribution container.