findHVG: Finding highly variable genes (HVG) based on Gini...

In jingshuw/descend: Gene Expression Distribution Deconvolution for Single Cell RNA-seq

Description

The function finds highly variable genes based on the Gini or CV estimates from DESCEND. A quantile natural cubic spline regression of the estimated selected criteria on the log of the estimated mean for the genes is performed as the general trend. HVGs are selected as the genes whose normalized difference between its estimates and the trend is larger than a threshold. For examples, see runDescend

Usage

findHVG(descend.list, criteria = c("Gini", "CV"), quantile = 0.7,
  threshold = 3, plot.result = T, spline.df = 5)

Arguments

descend.list	a list of descend objects computed from runDescend
criteria	the cirteria of HVG finding, either "Gini" or "CV"
quantile	quantile of the quantile regression. Defined as the tau parameter in the function rqss. A larger quantile yields a more stringent selection of HVG. Default is 0.7.
threshold	threshold of the normalized difference for HVG selection. Default is 3. A larger value results in a more stringent selection
plot.result	whether plot the selection results or not. Default is TRUE.
spline.df	the degree of freedom of the spline functions used to fit the quantile regression curve

Value

A list of elements:

score.mat	A score matrix of the genes. Each row is a gene
HVG.genes	A vector of the selected HVG gene names

jingshuw/descend documentation built on Nov. 2, 2021, 4:23 p.m.