scib_metrics.nmi_ari_cluster_labels_leiden

scib_metrics.nmi_ari_cluster_labels_leiden(X, labels, optimize_resolution=True, resolution=1.0, n_jobs=1)

Compute nmi and ari between leiden clusters and labels.

This deviates from the original implementation in scib by using leiden instead of louvain clustering. Installing joblib allows for parallelization of the leiden resoution optimization.

Parameters:

• X (NeighborsResults) – A NeighborsResults object.

• labels (ndarray) – Array of shape (n_cells,) representing label values

• optimize_resolution (bool (default: True)) – Whether to optimize the resolution parameter of leiden clustering by searching over 10 values

• resolution (float (default: 1.0)) – Resolution parameter of leiden clustering. Only used if optimize_resolution is False.

• n_jobs (int (default: 1)) – Number of jobs for parallelizing resolution optimization via joblib. If -1, all CPUs are used.

Return type:

dict[str, float]

Returns:

nmi

Normalized mutual information score

ari

Adjusted rand index score