scbiot.pl.plot_anndata_confusion

scbiot.pl.plot_anndata_confusion#

scbiot.pl.plot_anndata_confusion(adata, true_key, pred_key, *, dropna=True, drop_unknown=True, unknown_values=('unknown',), order_key=None, normalize='pred', figsize=(10, 3.6), cmap='Blues', linewidths=0.5, linecolor='lightgrey', annotate_mapping=True, mapping_fontsize=10, tick_fontsize=14, title=None, return_data=False)#

Plot a confusion-style heatmap comparing two categorical annotations in adata.obs.

Parameters#

adataAnnData-like

Must have adata.obs as a pandas DataFrame.

true_keystr

Column in adata.obs for ground-truth labels.

pred_keystr

Column in adata.obs for predicted labels.

dropnabool

Drop rows where either label is NA.

drop_unknownbool

Drop rows where either label matches values in unknown_values (case-insensitive).

unknown_valuesSequence[str]

Values treated as unknown.

order_keyOptional[str]

Which column to use for frequency-based ordering. Defaults to true_key.

normalizestr

“pred” => rows sum to 1 (fraction within predicted class). “true” => columns sum to 1 (fraction within true class). “all” => all entries sum to 1. “none” => raw counts.

figsize(w, h)

Figure size for the heatmap panel; when annotate_mapping is True, extra height is added.

cmapstr

Colormap name.

linewidths, linecolorfloat, str

Heatmap grid styling.

annotate_mappingbool

Show numeric ticks with side mapping text.

mapping_fontsize, tick_fontsizeint

Font sizes.

titleOptional[str]

Title for the heatmap.

return_databool

If True, also return (counts_df, normalized_df).

Returns#

(fig, ax) or (fig, ax, counts_df, norm_df)

Examples#

Basic usage:

>>> import scbiot as scb
>>> fig, ax = scb.pl.plot_anndata_confusion(adata, "true_cell_type", "pred_cell_type")
Parameters:

  • true_key (str)

  • pred_key (str)

  • dropna (bool)

  • drop_unknown (bool)

  • unknown_values (Sequence[str])

  • order_key (Optional[str])

  • normalize (str)

  • figsize (Tuple[float, float])

  • cmap (str)

  • linewidths (float)

  • linecolor (str)

  • annotate_mapping (bool)

  • mapping_fontsize (int)

  • tick_fontsize (int)

  • title (Optional[str])

  • return_data (bool)

Return type:

Union[Tuple[‘plt.Figure’, ‘plt.Axes’], Tuple[‘plt.Figure’, ‘plt.Axes’, pd.DataFrame, pd.DataFrame]]

Contents
Show Source