Get top highly variable genes.

Usage

getHVG(spe, n = 1000, min.total.count = 100, min.prop = 0.01)

Arguments

spe: A SpatialExperiment object.
n: Integer. The number of HVGs.
min.total.count: Numeric. Genes with total counts less than min.total.count across all cells are not considered for HVGs.
min.prop: Numeric. Genes that have non-zero counts in less than the specified proportion (min.prop) of all cells are excluded from HVG selection.

Value

A SpatialExperiment object with HVG information stored in rowData(spe)$hvg as a logical vector.

Details

getHVG adopts a fast approach of NB dispersion estimation for all the genes across all cells. A lowess curve is fit to represent mean-dispersion trend. Top HVGs are selected based on the ratio of gene-wise dispersion and their trended dispersion.

Examples


data("xenium_bc_spe")
spe <- getHVG(spe, n=100)