[BioC] population-by-environment interaction term in LIMMA

mrjmorri at ucalgary.ca mrjmorri at ucalgary.ca
Fri Sep 23 08:11:09 CEST 2011


Hello everyone, and thank you for taking the time to listen to my
 questions.  Its really great to know that there are people who are willing
 to share their experience, or at least helpfully point me in the right
 direction!

I have a simple experimental design, in which marine and freshwater fish
 were raised at two different temperatures.  I would like to be able to set
 up a model that is going to give me population, treatment, and population
 x treatment interaction terms for each gene.  Furthermore, if possible, I
 would like to include tank effect and sex as a variable in the model.  Any
 suggestions?  I see that lots of people generally do 2x2 factorial designs
 in which they only wish to contrast treatment within a line, and
interaction is never looked for.

The code I am currently using is as follows, but it obviously only gives
 me part of the story.  Guidance would be greatly appreciated!

normalize<-normalizeBetweenArrays(E, method="quantile")
dat1<-normalize[normalize$genes$ControlType==0,]
Eavg<-avereps(dat1, ID=dat1$genes$ProbeName)
TS <- paste(targets$Population, targets$Temperature, sep=".")
TS <- factor(TS, levels=c("marine.7","fw.7","marine.23","fw.23"))
 design <- model.matrix(~0+TS)
colnames(design) <- levels(TS)
fit <- lmFit(Eavg, design)
cont.matrix <- makeContrasts(marine.7vs23=marine.23-marine.7,
fw.7vs23=fw.23-fw.7)
fit2 <- contrasts.fit(fit, cont.matrix)
fit2 <- eBayes(fit2)
results <- decideTests(fit2, method="global")

Thanks!

Matthew



More information about the Bioconductor mailing list