past.Evaluation.run_leiden

past.Evaluation.run_leiden(adata, n_cluster, refine=False, num_nbs=6, use_rep='STAGATE', range_min=0, range_max=3, max_steps=30)

Search resolution so that leiden clustering algorithm obtain cluster numbers as close to given number as possible

Parameters:

• adata – target dataset of anndata format with latent feature stored in adata.obsm[use_rep] or with result of pp.neighbors

• n_cluster – cluster numbers

• refine – whether or not refine clustering results, if True, spatial coordinate should be stored in adata.obsm[“spatial”]

• num_nbs – number of neighbors to consider when refining clustering labels, valid only if refine is True

• use_rep – key of adata.obsm implying latent features

• range_min – start resolution to search

• range_max – end resolution to search

• max_steps – max iterators to search resolution

Returns:

target dataset of anndata format with searched leiden clustering result stored in adata.obs[“Nleiden”] and refined clustering result stored in adata.obs[“Nleiden_refined”] if refined is True

Return type:

scanpy.anndata