Add values_to_plot support to matrixplot and heatmap (issue #3842)

src/scanpy/plotting/__init__.py

@@ -110,3 +110,9 @@
 timeseries = deprecated("Use `dpt_timeseries`.")(timeseries)
 timeseries_as_heatmap = deprecated("Use `dpt_timeseries`.")(timeseries_as_heatmap)
 timeseries_subplot = deprecated("Use `dpt_timeseries`.")(timeseries_subplot)
+
+from ._rank_genes_groups import (
+    rank_genes_groups_dotplot,
+    rank_genes_groups_heatmap,
+    rank_genes_groups_matrixplot,
+)

src/scanpy/plotting/_rank_genes_groups.py

@@ -0,0 +1,167 @@
+from __future__ import annotations
+
+from typing import Literal, Sequence
+
+import numpy as np
+import pandas as pd
+
+from anndata import AnnData
+
+from ._anndata import heatmap, matrixplot, dotplot
+
+
+ArrayLike = np.ndarray
+ValuesToPlot = Literal["scores", "logfoldchanges", "pvals", "pvals_adj"]
+
+
+def _extract_rgg_values(
+    adata: AnnData,
+    values_to_plot: ValuesToPlot,
+    groups: Sequence[str] | str | None,
+    n_genes: int,
+):
+    """Extract dataframe (groups × genes) for the selected rank_genes_groups metric."""
+    rgg = adata.uns.get("rank_genes_groups", None)
+    if rgg is None:
+        raise ValueError("`adata.uns['rank_genes_groups']` not found.")
+
+    groups_order = rgg["names"].dtype.names
+    if isinstance(groups, str) and groups != "all":
+        groups = [groups]
+    elif groups is None or groups == "all":
+        groups = list(groups_order)
+
+    # gather top N genes per group
+    selected_genes = []
+    for g in groups:
+        arr = rgg["names"][g][:n_genes]
+        selected_genes.extend(arr)
+
+    selected_genes = list(dict.fromkeys(selected_genes))  # deduplicate, preserve order
+
+    # build dataframe values_df[group][gene]
+    df = pd.DataFrame(index=groups, columns=selected_genes, dtype=float)
+
+    for g in groups:
+        metrics = rgg[values_to_plot][g]  # ndarray length = total ranked genes
+        names = rgg["names"][g]
+
+        # map each gene to its metric
+        mapping = {gene: metrics[i] for i, gene in enumerate(names)}
+
+        # fill row
+        df.loc[g, :] = [mapping.get(gn, np.nan) for gn in selected_genes]
+
+    return df, selected_genes, groups
+
+
+# ------------------------------------------------------------------------------
+# Matrixplot
+# ------------------------------------------------------------------------------
+
+def rank_genes_groups_matrixplot(
+    adata: AnnData,
+    *,
+    values_to_plot: ValuesToPlot | None = None,
+    groups: Sequence[str] | str | None = None,
+    n_genes: int = 20,
+    **kwargs,
+    MatrixPlot wrapper for rank_genes_groups with DE metric selection.
+
+    Example:
+        sc.pl.rank_genes_groups_matrixplot(
+            adata,
+            values_to_plot="logfoldchanges",
+            groups=["0","1"],
+            n_genes=20,
+        )
+    """
+    if values_to_plot is None:
+        # default: plot expression of marker genes
+
+        raise ValueError(
+            "`values_to_plot` must be provided. Options: "
+            "['scores', 'logfoldchanges', 'pvals', 'pvals_adj']"
+        )
+
+    values_df, genes, groups = _extract_rgg_values(
+        adata, values_to_plot, groups, n_genes
+    )
+
+    return matrixplot(
+        adata,
+        var_names=genes,
+        groupby=groups,
+        values_df=values_df,
+        **kwargs,
+    )
+
+
+# ------------------------------------------------------------------------------
+# Heatmap
+# ------------------------------------------------------------------------------
+
+def rank_genes_groups_heatmap(
+    adata: AnnData,
+    *,
+    values_to_plot: ValuesToPlot | None = None,
+    groups: Sequence[str] | str | None = None,
+    n_genes: int = 20,
+    **kwargs,
+):
+    """
+    Heatmap wrapper for rank_genes_groups with DE metric selection.
+    """
+    if values_to_plot is None:
+        raise ValueError(
+            "`values_to_plot` must be provided. Options: "
+            "['scores', 'logfoldchanges', 'pvals', 'pvals_adj']"
+        )
+
+    values_df, genes, groups = _extract_rgg_values(
+        adata, values_to_plot, groups, n_genes
+    )
+
+    return heatmap(
+        adata,
+        var_names=genes,
+        groupby=groups,
+        values_df=values_df,
+        **kwargs,
+    )
+
+
+# ------------------------------------------------------------------------------
+# Dotplot (for completeness parity with the issue text)
+# ------------------------------------------------------------------------------
+
+def rank_genes_groups_dotplot(
+    adata: AnnData,
+    *,
+    values_to_plot: ValuesToPlot | None = None,
+    groups: Sequence[str] | str | None = None,
+    n_genes: int = 20,
+    **kwargs,
+):
+    """
+    DotPlot wrapper for rank_genes_groups with DE metric selection.
+    This adds parity with the existing sc.pl.rank_genes_groups_dotplot API.
+    """
+    if values_to_plot is None:
+        raise ValueError(
+            "`values_to_plot` must be provided. Options: "
+            "['scores', 'logfoldchanges', 'pvals', 'pvals_adj']"
+        )
+
+    values_df, genes, groups = _extract_rgg_values(
+        adata, values_to_plot, groups, n_genes
+    )
+
+    # DotPlot uses values_df as dot_color_df
+    return dotplot(
+        adata,
+        var_names=genes,
+        groupby=groups,
+        dot_color_df=values_df,
+        **kwargs,
+    )