came.aligned_datapair_from_adatas¶

came.aligned_datapair_from_adatas(adatas: Sequence[AnnData], vars_feat: Sequence, vars_as_nodes: Sequence | None = None, oo_adjs: Sequence[spmatrix] | None = None, dataset_names: Sequence[str] = ('reference', 'query'), **kwds) → AlignedDataPair¶

Build AlignedDataPair object from a pair of adatas.

Note that the node features will be extracted from adata.raw (if not None), so please make sure that these values are normalized.

Parameters:

adatas – a list or tuple of 2 sc.AnnData objects.
vars_feat – a list of variable-names that will be used as (cell) node features. for example, names of differentail expressed genes (DEGs), highly variable features.
vars_as_nodes – a sequence of variable names, optional. a name-list of variables that will be taken as nodes in the graph for model training. if None (not provided), it will be the same as vars_feat
oo_adjs – a sequence of (sparse) adjacent matrices of observations. for example, the single-cell network within each dataset.
dataset_names – list or tuple of 2. names to discriminate data source, e.g. (‘reference’, ‘query’)

Returns:

dpair

Return type:

AlignedDataPair

Examples

>>> dpair = aligned_datapair_from_adatas(
...     [adata1, adata2],
...     vars_feat,
...     dataset_names = ['reference', 'query']
...     )