omicverse.pp.louvain

Contents

omicverse.pp.louvain#

omicverse.pp.louvain(adata, resolution=None, *, random_state=0, key_added='louvain', neighbors_key=None, directed=True, use_weights=False, copy=False, **kwargs)[source]#

Run Louvain clustering on the precomputed kNN graph.

Parameters:

  • adata (anndata.AnnData) – AnnData object with a neighborhood graph already computed.

  • resolution (default: None) – Resolution γ; higher values yield more, smaller clusters.

  • random_state (default: 0) – Seed for reproducibility.

  • key_added (default: 'louvain') – Column in adata.obs to store labels. Default 'louvain'.

  • neighbors_key (default: None) – Key in adata.uns for the neighbor dict. None uses the standard 'neighbors' slot.

  • directed (default: True) – Forwarded to the igraph backend; rarely need tuning.

  • use_weights (default: False) – Forwarded to the igraph backend; rarely need tuning.

  • copy (default: False) – Return a new AnnData instead of modifying in place.

  • **kwargs – Forwarded to the underlying backend (scanpy.tl.louvain or rapids_singlecell.tl.louvain).

Notes

In ov.settings.mode='cpu-gpu-mixed' this is auto-routed to ov.pp.leiden (Louvain has no GPU backend and Leiden is its strict improvement). Labels are still written to adata.obs[key_added] so existing notebooks keep working.

Contents