grassp.pl.tagm_map_contours

tagm_map_contours(adata, embedding='umap', size=100, components=None, dimensions=None, levels=4, ax=None, **kwargs)

Plot posterior density contours of TAGM components in embedding space.

The function draws synthetic observations from the TAGM posterior via sample_tagm_map() and projects them into either UMAP or PCA space (depending on embedding). A kernel–density estimate is then computed for each component and visualized as contour lines.

Parameters:

adata AnnData

Annotated data matrix that already contains TAGM posterior parameters in adata.uns["tagm.map.params"] and an embedding (UMAP or PCA) fitted by umap() or scanpy.pp.pca().

embedding Literal['umap', 'pca'] (default: 'umap')

Target space for the contours, either "umap" or "pca".

size int (default: 100)

Number of posterior samples to draw per component.

components Union[str, Sequence[str], None] (default: None)

Specify which dimensions of the embedding to use. Only one pair of dimensions is allowed. Use components (Scanpy style string) or a dimensions tuple, not both.

dimensions Union[tuple[int, int], Sequence[tuple[int, int]], None] (default: None)

Specify which dimensions of the embedding to use. Only one pair of dimensions is allowed. Use components (Scanpy style string) or a dimensions tuple, not both.

levels int (default: 4)

Number of contour levels passed to seaborn.kdeplot().

ax Axes | None (default: None)

Matplotlib axes to plot on. If None, the current axes returned by matplotlib.pyplot.gca() are used.

**kwargs

Additional keyword arguments forwarded to seaborn.kdeplot().

Return type:

Axes

Returns:

The axes object containing the contour plot (returned for convenience).