get_model_var: Return average variance under negative binomial model

In sctransform: Variance Stabilizing Transformations for Single Cell UMI Data

Return average variance under negative binomial model

Description

This is based on the formula var = mu + mu^2 / theta

Usage

get_model_var(
  vst_out,
  cell_attr = vst_out$cell_attr,
  use_nonreg = FALSE,
  bin_size = 256,
  verbosity = 2
)

Arguments

vst_out	The output of a vst run
cell_attr	Data frame of cell meta data
use_nonreg	Use the non-regularized parameter estimates; boolean; default is FALSE
bin_size	Number of genes to put in each bin (to show progress)
verbosity	An integer specifying the verbosity level: 0 (silent, no messages), 1 (show messages only), or 2 (show messages and progress bars); default is 2

Value

A named vector of variances (the average across all cells), one entry per gene.

Examples

vst_out <- vst(pbmc, return_cell_attr = TRUE)
res_var <- get_model_var(vst_out)

sctransform documentation built on Jan. 10, 2026, 9:17 a.m.