snapatac2.tl.aggregate_X#

snapatac2.tl.aggregate_X(adata, groupby=None, normalize=None, file=None)[source]#

Aggregate .X values across cells or cell groups.

Use this function to create pseudobulk count profiles, optionally normalized as RPM or RPKM, from all cells or from groups defined by groupby.

Anti-Patterns#

  • Do NOT use normalize="RPKM" unless adata.var_names are genomic regions in chrom:start-end format.

  • Do NOT expect a raw NumPy array return; the function always returns an AnnData object containing the aggregated matrix.

param adata:

Annotated data object with cells in observations and features in variables.

type adata:

AnnData | AnnDataSet

param groupby:

Grouping key in adata.obs, one group label per cell, or None to aggregate all cells together.

type groupby:

str | list[str] | None

param normalize:

Optional normalization applied to each aggregated profile.

type normalize:

Optional[Literal['RPM', 'RPKM']]

param file:

Output h5ad path for backed results. If None, return an in-memory AnnData.

type file:

Path | None

returns:

AnnData object with aggregated profiles in .X, original feature names in .var_names, and group names in .obs_names when groupby is set.

rtype:

AnnData

Examples

>>> import snapatac2 as snap
>>> adata = snap.datasets.pbmc5k(type="annotated_h5ad")
>>> pseudobulk = snap.tl.aggregate_X(adata, groupby="cell_type", normalize="RPM")
>>> pseudobulk.n_vars == adata.n_vars
True
On this page