grassp.tl.aligned_umap

aligned_umap(data_list, align_data=True, return_data_objects=True, n_neighbors=20, metric='euclidean', min_dist=0.1, alignment_regularisation=0.002, n_epochs=300, random_state=None, verbose=False, n_components=2)

Aligned UMAP embedding for matched datasets.

Wraps umap.AlignedUMAP and writes the resulting coordinates to .obsm['X_aligned_umap'] of each input object.

Parameters:

data_list List[AnnData]

List of AnnData objects assumed to contain the same observations and variables (will be intersected if align_data is True).

align_data bool (default: True)

Whether to call align_adatas() before fitting.

return_data_objects bool (default: True)

If True (default) return the modified list; otherwise return the fitted umap.AlignedUMAP estimator.

n_neighbors int (default: 20)

Number of nearest neighbours to construct the UMAP graph.

metric str (default: 'euclidean')

Distance metric used by UMAP (e.g. "euclidean").

min_dist float (default: 0.1)

Minimum separation of points in the low-dimensional space.

alignment_regularisation float (default: 0.002)

Weight of the between-dataset alignment objective.

n_epochs int (default: 300)

Training epochs.

n_components int (default: 2)

Dimensionality of the embedding (usually 2).

random_state int | None (default: None)

Passed directly to umap.AlignedUMAP.

verbose bool (default: False)

Passed directly to umap.AlignedUMAP.

Return type:

AlignedUMAP

Returns:

list[AnnData] | umap.AlignedUMAP Depending on return_data_objects.