contrasts.varFit: Compute contrasts for a varFit object.
In ChuanJ/test: Analysing Illumina HumanMethylation BeadChip Data

Description Usage Arguments Details Value Author(s) See Also Examples

Compute estimated coefficients, standard errors and LogVarRatios for a given set of contrasts.

1	contrasts.varFit(fit, contrasts=NULL)

`fit`	list containing a linear model fit produced by `varFit`. The fit object should be of class `MArrayLM`.
`contrasts`	numeric matrix with rows corresponding to coefficients in `fit` and columns containing contrasts.

This function calls the contrasts.fit function in limma to compute coefficients and standard errors for the specified contrasts corresponding to a linear model fit obtained from the varFit function. LogVarRatios are also computed in terms of the contrasts. A contrasts matrix can be computed using the makeContrasts function.

A list object of the same class as fit.

Belinda Phipson

varFit, contrasts.fit, makeContrasts

# Randomly generate data for a 3 group problem with 100 CpG sites and 4 arrays in each group. 

library(limma)

y<-matrix(rnorm(1200),ncol=12)

group<-factor(rep(c(1,2,3),each=4))
design<-model.matrix(~0+group)
colnames(design)<-c("grp1","grp2","grp3")

# Fit linear model for differential variability
vfit<-varFit(y,design)

# Specify contrasts
contr<-makeContrasts(grp2-grp1,grp3-grp1,grp3-grp2,levels=colnames(design))

# Compute contrasts from fit object
vfit.contr<-contrasts.varFit(vfit,contrasts=contr)

summary(decideTests(vfit.contr))

# Look at top table of results for first contrast

topVar(vfit.contr,coef=1)