troutpy.tl.spatial_variability

troutpy.tl.spatial_variability#

troutpy.tl.spatial_variability(sdata, coord_keys=None, gene_key='gene', n_neighbors=10, kde_resolution=1000, square_size=20, n_threads=1, method='moran', copy=False)#

Compute spatial variability of extracellular RNA using Moran’s I (or another autocorrelation statistic).

Extracellular transcripts are binned onto a spatial grid with LazyKDE, and spatial autocorrelation is computed per gene on the resulting grid using squidpy. Results are stored in sdata["xrna_metadata"].var.

Parameters:

  • sdata (spatialdata.SpatialData) – SpatialData object with a "transcripts" points layer containing coord_keys, "extracellular", and gene_key columns.

  • coord_keys (list of str, optional) – Spatial coordinate column names. Defaults to ["x", "y"].

  • gene_key (str, optional) – Column in the transcript layer with gene identifiers. Defaults to "gene".

  • n_neighbors (int, optional) – Number of neighbors used to build the spatial neighbor graph. Defaults to 10.

  • kde_resolution (int, optional) – Grid resolution passed to LazyKDE. Defaults to 1000.

  • square_size (int, optional) – Bin size (in coordinate units) for the KDE grid. Defaults to 20.

  • n_threads (int, optional) – Number of threads for LazyKDE processing. Defaults to 1.

  • method (str, optional) – Spatial autocorrelation statistic passed to squidpy.gr.spatial_autocorr(). Defaults to "moran".

  • copy (bool, optional) – If True, return the updated SpatialData object; otherwise modify in place and return None. Defaults to False.

Returns:

spatialdata.SpatialData or None Updated SpatialData with f"{method}_*" columns added to sdata["xrna_metadata"].var if copy=True; otherwise None.

Contents