scanpy.tl.diffmap

scanpy.tl.diffmap(adata, n_comps=15, *, neighbors_key=None, random_state=0, copy=False)

Diffusion Maps [Coifman et al., 2005, Haghverdi et al., 2015, Wolf et al., 2018].

Diffusion maps [Coifman et al., 2005] has been proposed for visualizing single-cell data by Haghverdi et al. [2015]. The tool uses the adapted Gaussian kernel suggested by Haghverdi et al. [2016] in the implementation of Wolf et al. [2018].

The width (“sigma”) of the connectivity kernel is implicitly determined by the number of neighbors used to compute the single-cell graph in neighbors(). To reproduce the original implementation using a Gaussian kernel, use method=='gauss' in neighbors(). To use an exponential kernel, use the default method=='umap'. Differences between these options shouldn’t usually be dramatic.

Parameters:

adata AnnData

Annotated data matrix.

n_comps int (default: 15)

The number of dimensions of the representation.

neighbors_key str | None (default: None)

If not specified, diffmap looks .uns[‘neighbors’] for neighbors settings and .obsp[‘connectivities’], .obsp[‘distances’] for connectivities and distances respectively (default storage places for pp.neighbors). If specified, diffmap looks .uns[neighbors_key] for neighbors settings and .obsp[.uns[neighbors_key][‘connectivities_key’]], .obsp[.uns[neighbors_key][‘distances_key’]] for connectivities and distances respectively.

random_state int | RandomState | None (default: 0)

A numpy random seed

copy bool (default: False)

Return a copy instead of writing to adata.

Return type:

AnnData | None

Returns:

Returns None if copy=False, else returns an AnnData object. Sets the following fields:

adata.obsm['X_diffmap']numpy.ndarray (dtype float)

Diffusion map representation of data, which is the right eigen basis of the transition matrix with eigenvectors as columns.

adata.uns['diffmap_evals']numpy.ndarray (dtype float)

Array of size (number of eigen vectors). Eigenvalues of transition matrix.

Notes

The 0-th column in adata.obsm["X_diffmap"] is the steady-state solution, which is non-informative in diffusion maps. Therefore, the first diffusion component is at index 1, e.g. adata.obsm["X_diffmap"][:,1]