grassp.tl.silhouette_score

silhouette_score(data, gt_col, use_rep='X_umap', key_added='silhouette', inplace=True)

Per-group silhouette scores.

Computes the silhouette score for each group in data.obs[gt_col].

Parameters:

data

AnnData object containing an embedding in .obsm.

gt_col

Column in data.obs with cluster labels.

use_rep default: 'X_umap'

Key of the embedding to evaluate (default "X_umap").

key_added default: 'silhouette'

Base key under which results are stored (default "silhouette").

inplace default: True

If True (default) store return None, if False return the silhouette scores.

Return type:

None | ndarray

Returns:

If inplace is True:

data.obs[key_added]

Vector of silhouette scores.

data.uns[key_added]['mean_silhouette_score']

Global mean.

data.uns[key_added]['cluster_mean_silhouette']

Mapping of cluster → mean score.

If inplace is False:

Vector of silhouette scores.