genome_spy.anndata.from_anndata

genome_spy.anndata.from_anndata(adata, *, obs=None, genes=(), obsm_keys=())View on GitHub

Prepare observation metadata, selected genes, and embedding components.

Description:

Returns a pandas snapshot with one row per observation in source order. Gene fields use exact var_names and read X; embedding fields use Scanpy’s <key>-<component> naming. Only selected expression columns are densified. No analysis, file loading, or AnnData mutation occurs. The index contains observation names; charts omit it unless explicitly exported with pandas reset_index().

Parameters:

  • adata – Source AnnData object or view.

  • obs – Observation columns to include. None includes all; [] includes none.

  • genes – Exact variable names to extract from X.

  • obsm_keys – Embedding keys and zero-based integer component indices.

Returns:

A pandas table accepted by Chart, Arrow export, and widget updates.

Raises:

  • ImportError – If AnnData is unavailable; install the anndata extra.

  • TypeError – If the source is not AnnData or selected arrays are neither NumPy arrays, SciPy sparse arrays/matrices, nor pandas embeddings.

  • AnnDataProjectionError – If fields are absent or ambiguous, components are invalid, or requested output names collide.

Example

>>> import genome_spy as gs
>>> from genome_spy.anndata import from_anndata
>>> cells = from_anndata(
...     adata, obs=["cell_type"], genes=["MS4A1"],
...     obsm_keys=[("X_umap", 0), ("X_umap", 1)],
... )
>>> chart = gs.Chart(cells).mark_point().encode(
...     x="X_umap-0", y="X_umap-1", color="MS4A1"
... )