grassp.tl.to_knn_graph

to_knn_graph(data, node_label_column=None, neighbors_key=None, obsp=None)

Convert the k-NN graph stored in AnnData to a networkx graph.

Parameters:

data AnnData

anndata.AnnData that has been processed with scanpy.pp.neighbors() (or equivalent) so that either adata.obsp[obsp] or adata.uns[neighbors_key] exists.

node_label_column str | None (default: None)

Observation column whose values become node labels in the resulting graph. If None (default), data.obs_names is used.

neighbors_key str | None (default: None)

Key under which Scanpy stored neighbour information (defaults to 'neighbors'). Ignored if obsp is provided.

obsp str | None (default: None)

Name of a pre-computed adjacency/connectivity matrix in adata.obsp. Takes precedence over neighbors_key.

Return type:

Graph

Returns:

networkx.Graph Undirected, weighted graph where edge weights correspond to the connectivities/distances of the k-NN graph.