pcrplot: Principal component regression plot

In xuz1/ENmix: Quality control and analysis tools for Illumina DNA methylation BeadChip

Principal component regression plot

Description

First, principal component analysis will be performed in the standadized input data matrix (standadized for each row/CpG), and then the specified number of top principal components (which explain most data variation) will be used to perform linear regression with each specified variable. Regression P values will be plotted for exploration of methylation data variance structure or identification of possible confounding variables in association analysis.

Usage

pcrplot(beta, cov,npc=50)

Arguments

beta	A methylation beta value matrix with rows for probes and columns for samples. The input matrix should not contain any missing value.
cov	A data frame of covariates. Categorical variables should be converted to factors. The number of rows should equal to the number of columns in beta matrix
npc	The number of top ranked principal components to be plotted

Value

A jpeg figure "svdscreeplot.jpg" to show the variations explained by each principal component.

A jpeg figure "pcr_diag.jpg" to show association strength between principal components and covariates with cell colors indicating different levels of association P values.

Author(s)

Zongli Xu

References

Zongli Xu, Liang Niu, Leping Li and Jack A. Taylor, ENmix: a novel background correction method for Illumina HumanMethylation450 BeadChip. Nucleic Acids Research 2015

Examples

if (require(minfiData)) {
mdat <- preprocessRaw(RGsetEx)
beta=getBeta(mdat, "Illumina")
group=pData(mdat)$Sample_Group
slide=factor(pData(mdat)$Slide)
cov=data.frame(group,slide)
pcrplot(beta,cov,npc=6)
}

xuz1/ENmix documentation built on Nov. 24, 2024, 4:31 a.m.