Perform kernel density estimation on SpatialExperiment for cell types of interest

Usage

gridDensity(
  spe,
  id = if (isVisium) "in_tissue" else "cell_type",
  coi = NULL,
  kernel = "gaussian",
  bandwidth = NULL,
  ngrid.x = NULL,
  grid.length.x = NULL,
  diggle = FALSE,
  grid.type = c("hex", "square"),
  isVisium = FALSE,
  filterToVisiumSpot = isVisium
)

Arguments

spe: A SpatialExperiment object.
id: A character. The name of the column of colData(spe) containing the cell type identifiers. Set to cell_type by default.
coi: A character vector of cell types of interest (COIs). Default to all cell types.
kernel: The smoothing kernel. Options are "gaussian", "epanechnikov", "quartic" or "disc". For hexagonal grid, only Gaussian is implemented
bandwidth: The smoothing bandwidth. By default performing automatic bandwidth selection using cross-validation using function spatstat.explore::bw.diggle.
ngrid.x: Number of grids in the x-direction. Ignored when 'grid.length.x' is specified. Default to NULL.
grid.length.x: Grid length in the x-direction. If both 'ngrid.x' and 'grid.length.x' are NULL, then 'grid.length.x' is set to 100 (micron) by default.
diggle: Logical. If TRUE, use the Jones-Diggle improved edge correction. See spatstat.explore::density.ppp() for details.
grid.type: Type of grid can be either hexagon or square.
isVisium: Logical. If TRUE, fit hexagonal grids to Visium spots by replacing spatial coords with array rows & array cols.
filterToVisiumSpot: Logical. If TRUE, filter grid polygons to only those with a Visium spot underneath.

Value

A SpatialExperiment object. Grid density estimates for all cell type of interest are stored in spe@metadata$grid_density. Grid information is stored in spe@metadata$grid_info

Examples


data("xenium_bc_spe")

spe <- gridDensity(spe)