Meta cells highly variable genes — rs_mc

Calculates highly variable genes for MetaCells or more generally speaking sparse data. This is happening in-memory compared to the (usually much) larger single cell data sets. "meanvarbin" and "dispersion" compute the same statistics; they differ only in how the R side selects from them.

Usage

rs_mc_hvg(
  sparse_data,
  hvg_method,
  loess_span,
  binning,
  n_bins,
  clip_max,
  verbose
)

Arguments

sparse_data: A named list that needs to have data, indptr, indices, nrow, ncol and cs_type. Shape is (metacells, genes). Pass raw counts for "vst" and normalised counts otherwise.
hvg_method: String. Which HVG detection method to use. Options are c("vst", "meanvarbin", "dispersion").
loess_span: Numeric. The span parameter for the loess function (only used for "vst").
binning: String. The binning strategy for the meanvarbin and dispersion methods. One of c("equal_width", "equal_freq").
n_bins: Integer. Number of bins for the meanvarbin and dispersion methods.
clip_max: Optional clipping number. Defaults to sqrt(no_cells) if not provided (only used for "vst").
verbose: Integer. 0L - quiet; 1L - normal verbosity; 2L - detailed verbosity.

Value

A list with the HVG statistics. If hvg_method == "vst":

mean - The average expression of the gene.
var - The variance of the gene.
var_exp - The expected variance of the gene.
var_std - The standardised variance of the gene.

For "meanvarbin" and "dispersion":

mean - The average expression of the gene.
dispersion - The dispersion of the gene.
dispersion_scaled - The scaled dispersion per bin per gene.
bin - The bin of the gene.