Wrapper function for HVG detection parameters.

Usage

params_sc_hvg(
  method = c("vst", "meanvarbin", "dispersion", "residual"),
  loess_span = 0.3,
  num_bin = 20L,
  bin_method = c("equal_width", "equal_freq")
)

Arguments

method: String. "residual" ranks genes by the residual variance of a model fitted with fit_residuals_sc(), and needs that fit on the object first. It also treats hvg_no as a per-group count and returns the union across groups, so a grouped fit can select more than hvg_no genes. One of c("vst", "meanvarbin", "dispersion", "residual"). Defaults to "vst".
loess_span: Numeric. The span parameter for the loess function that is used to standardise the variance for method = "vst". Defaults to 0.3.
num_bin: Integer. Not yet implemented. Defaults to 20L.
bin_method: String. The binning method. One of c("equal_width", "equal_freq"). Defaults to "equal_width".

Value

A named list with the following elements:

method - String. "residual" ranks genes by the residual variance of a model fitted with fit_residuals_sc(), and needs that fit on the object first. It also treats hvg_no as a per-group count and returns the union across groups, so a grouped fit can select more than hvg_no genes. One of c("vst", "meanvarbin", "dispersion", "residual"). Defaults to "vst".
loess_span - Numeric. The span parameter for the loess function that is used to standardise the variance for method = "vst". Defaults to 0.3.
num_bin - Integer. Not yet implemented. Defaults to 20L.
bin_method - String. The binning method. One of c("equal_width", "equal_freq"). Defaults to "equal_width".