Compute Local Cell Type Composition in the Spatial Neighborhood

Computes the local cell type composition for each spot or cell based on spatial proximity. Supports both single-cell resolution using nearest neighbors and spot-level resolution using distance thresholding with cell type proportions.

Usage

Cell_type_neighborhood(
  sample_information_coordinate,
  resolution = c("single_cell", "spot"),
  sample_information_cellType = NULL,
  sample_information_decon = NULL,
  k = 50,
  r = 4
)

Arguments

sample_information_coordinate

A data.frame or matrix of spatial coordinates (x, y). Row names must correspond to cell or spot IDs.

resolution

One of "single_cell" or "spot".

sample_information_cellType

Named vector of cell types for single-cell resolution. Names must match rownames of sample_information_coordinate.

sample_information_decon

A matrix of cell type proportions per spot (for "spot" mode). Row names must match those of sample_information_coordinate.

k

Integer, number of nearest neighbors (used only in "single_cell" mode). Default is 50.

r

Numeric, distance threshold for neighborhood (used only in "spot" mode). Default is 4.

Value

A matrix of normalized cell type proportions for each spatial unit (rows) across cell types (columns).

Details

This function calculates the distribution of cell types around each spatial unit using one of the following modes:

• "single_cell": Uses discrete cell-type labels and K-nearest neighbors.

• "spot": Uses continuous deconvolution proportions and radius threshold.

The returned matrix represents normalized cell-type compositions in each local neighborhood.

Author

Bin Duan

Examples

if (FALSE) { # \dontrun{
data("osmFISH_metadata_cellType")
data("osmFISH_bulk_decon")
data("osmFISH_bulk_pheno")
data("osmFISH_bulk_coordinate")

Cell_type_neighborhood(
  sample_information_coordinate = osmFISH_bulk_coordinate,
  resolution = "single_cell",
  sample_information_cellType = osmFISH_metadata_cellType
)

} # }