rapids_singlecell.gr.spatial_autocorr

rapids_singlecell.gr.spatial_autocorr(adata, *, connectivity_key='spatial_connectivities', genes=None, mode='moran', transformation=True, n_perms=None, two_tailed=False, corr_method='fdr_bh', layer=None, use_raw=False, use_sparse=True, copy=False)

Calculate spatial autocorrelation for genes in an AnnData object.

This function computes spatial autocorrelation scores (Moran’s I or Geary’s C) for each gene in an AnnData object. The function also calculates p-values and corrected p-values for multiple testing.

Note:

This implementation uses single-precision (float32) for calculations, which may result in decreased accuracy for weak correlations when compared to double-precision (float64) calculations. For strongly correlated data, the difference in p-values should be minimal. However, for weakly correlated data with I or C values close to their expected values, the lack of precision may lead to larger discrepancies in p-values.

Parameters:

adata AnnData

Annotated data matrix.

connectivity_key str (default: 'spatial_connectivities')

Key of the connectivity matrix in adata.obsp, by default “spatial_connectivities”.

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

Genes for which to compute the autocorrelation scores. If None, all genes or highly variable genes will be used.

mode Literal['moran', 'geary'] (default: 'moran')

Spatial autocorrelation method to use, either “moran” or “geary”, by default “moran”.

transformation bool (default: True)

If True, row-normalize the connectivity matrix, by default True.

n_perms int | None (default: None)

Number of permutations for calculating p-values, by default None.

two_tailed bool (default: False)

If True, calculate two-tailed p-values, by default False.

corr_method str | None (default: 'fdr_bh')

Multiple testing correction method to use, by default “fdr_bh”.

layer str | None (default: None)

Layer in the AnnData object to use, by default None.

use_raw bool (default: False)

If True, use the raw data in the AnnData object, by default False.

use_sparse bool (default: True)

If True, use a sparse representation for the input matrix vals when it is a sparse matrix, by default True.

copy bool (default: False)

If True, return the results as a DataFrame instead of storing them in adata.uns, by default False.

Return type:

DataFrame | None

Returns:

DataFrame containing the autocorrelation scores, p-values, and corrected p-values for each gene. If copy is False, the results are stored in adata.uns and None is returned.