prcompExprs: Principal component analysis of an expression matrix

In bedapub/ribiosNGS: Utilities for next-generation sequencing data analysis in ribios

Principal component analysis of an expression matrix

Description

Principal component analysis of an expression matrix

Usage

prcompExprs(matrix, ntop = NULL, scale = FALSE, nbin = NULL)

Arguments

matrix	Numeric matrix. Features in rows and samples in columns.
ntop	Integer or NULL. If not NULL, only ntop genes with the highest variance are used for the calculation.
scale	Logical, whether variance of features should be scaled to 1. Default FALSE, as recommended by Nguyen et al. (2019)
nbin	Integer. Genes are divided into nbin bins by their average gene expression signal, and top variable genes (approximately ntop/nbin) are selected from each bin. If NULL or NA, an automatic value (100, or nrow(matrix) %/% 10 when fewer are 1000 genes are used as input) is used. It is only used when ntop is not NULL.

References

Nguyen, Lan Huong, and Susan Holmes. "Ten Quick Tips for Effective Dimensionality Reduction." PLOS Computational Biology 15, no. 6 (2019): e1006907

See Also

topVarRowsByMeanBinning

Examples

myTestExprs <- matrix(rnorm(1000), ncol=10, byrow=FALSE)
myTestExprs[1:50, 6:10] <- myTestExprs[1:50, 6:10] + 2
myTopPca <- prcompExprs(myTestExprs, ntop=50, nbin=5)

bedapub/ribiosNGS documentation built on Feb. 10, 2025, 12:34 a.m.