Preprocessing Agilent Two-Color

Question

[babak1146]

I have an Agilent 2-Color dataset (GSE23611), i am trying to normalize it with Limma package. when I normalize dataset , hclust cannot distinguish between case and control after normalization. my code is :

library(limma)
files_case <- dir(pattern="*\\.txt$")
dat <- read.maimages(files_case,source="agilent")
dat2<-normalizeWithinArrays(dat, method="loess","normexp", offset=50)
dat2 <- normalizeBetweenArrays(dat2, method="Aquantile")
dat2 <- avereps(dat2, ID=dat2$genes$ProbeName)
dat.m<-dat2$M
rownames(dat.m)<-dat2$genes$ProbeName
dat.m=na.omit(dat.m)
dat.dist<-dist(t(dat.m))
plot(hclust(dat.dist))

is it true?

[Ram]

Please use the formatting bar (especially the code option) to present your post better. I've done it for you this time.

[Kevin Blighe]

Why were you expecting it to segregate cases and controls? The type of clustering that you are doing is unsupervised, i.e., using all probe expression values; thus, it should never be expected that this will segregate your dataset as you desire.

If you perform differential expression analysis between cases and controls and then perform supervised clustering using the statistically significant probes, then you should see segregation.