interscale.evaluation.scale_cls_by_sample

interscale.evaluation.scale_cls_by_sample(adata, sample_key, cls_columns=['combined_cls_horizontal', 'combined_cls_vertical'], inplace=True, suffix='_scaled')

Scale CLS token values to [0, 1] within each sample/window to make them comparable across windows of different sizes.

Parameters:

• adata – Annotated data object.

• sample_key (str) – Column name in adata.obs containing the sample/window identifiers.

• cls_columns (list (default: ['combined_cls_horizontal', 'combined_cls_vertical'])) – Column name(s) of CLS values to scale (e.g., "combined_cls_horizontal").

• inplace (bool (default: True)) – If True, add scaled columns to adata.obs with suffix; if False, return a DataFrame with scaled values.

• suffix (str (default: '_scaled')) – Suffix to add to column names when inplace=True.

Returns:

DataFrame with scaled values if inplace=False, otherwise None.