Calculate local Moran for 1 to 2 variables.

Usage

localMoran(
  spe,
  data1,
  data2 = data1[],
  at = c("grid", "cell"),
  nbrs_name = NULL,
  hhonly = FALSE,
  significance_cutoff = 0.05,
  permutations = 999,
  seed = 123456789,
  cpu_threads = 6
)

Arguments

spe: A SpatialExperiment object.
data1: Numeric vector 1. Must be same length as number of grid points or cells, depending on 'at'.
data2: Numeric vector 2 for bivariate local Moran. Must be same length as data1.
at: Option of grid or cell for where to look for neighbour list
nbrs_name: Name of the neighbour list in spe@metadata$grid[[at]] for Moran's I. If NULL, will use the newest neighbour list.
hhonly: Only high-high clusters, which is more interpretable. Other clusters (e.g. "Low-Low","High-Low",...) will be assigned as undefined.
significance_cutoff: Cutoff for p-value to filter non-significant clusters
permutations: Number of permutations for p-value.
seed: Integer. For random permutations.
cpu_threads: The number of cpu threads used for parallel computation.

Value

List of lisa_value, clusters, and pseudo p-value.

Examples

data("xenium_bc_spe")

## At grid.
spe <- gridDensity(spe, coi = "Breast cancer")
dat <- spe@metadata$grid_density$density_breast_cancer
spe <- findNbrsGrid(spe)
res <- localMoran(spe,data1=dat, at = "grid")

## At cell.
dat <- as.numeric(spe$cell_type=="Breast cancer")
spe <- findNbrsSpatial(spe,k=20)
res <- localMoran(spe,data1=dat,at="cell")