[BioC] limma and paired data

[Gordon K Smyth]

Why not follow the section in the limma User's Guide on technical replication?  Blocking on
individuals and technical replication is really the same thing.

Gordon

> Date: Mon, 21 Feb 2005 11:41:53 +0100
> From: <SKALKO at clinic.ub.es>
> Subject: [BioC] limma and paired data
> To: <bioconductor at stat.math.ethz.ch>
>
> Dear all,
>
>
>
> I have a question on a subject that I think was not discussed here
> before.
>
> I am using limma package for the detection of significant differential
> expression in an affy experiment:
>
>
>
> 3 "Healthy" (group1) indiv. before treatment and the same indiv. after
> treatment (group4)
>
> 3 "ill-low"   (group2)    "                "           "
> "           "                  (group 5)
>
> 2 "ill-high"  (group3)   "                 "           "
> "           "                  (group 6)
>
>
>
> the interest is comparing effects of the treatment (i.e. group4-group1,
> group5-group2, etc). I used these commands:
>
>
>
>>library(affy)
>
>>library(limma)
>
>>library("hgu133a")
>
>>x<-ReadAffy()
>
>
>
>>eset<-rma(x)
>
>>design <- model.matrix(~ -1+factor(c(1,1,1,2,2,2,3,3,4,4,4,5,5,5,6,6)))
>
>>colnames(design) <-
> c("group1","group2","group3","group4","group5","group6")
>
>
>
>>fit<-lmFit(eset,design)
>
>
>
>>contrast.matrix <-
> makeContrasts(group2-group1,group4-group1,group5-group2,
> group6-group3,levels=design)
>
>
>
>>fit2 <- contrasts.fit(fit, contrast.matrix)
>
>>fit2 <- eBayes(fit2)
>
>
>
> The question is:  How has to be taken into account that the individuals
> are the same before and after the treatment?
>
> I red about block in lmFit but I am not sure how to do that. Here it
> would be some correlation, but no so high as
>
> in the case of  real technical replicates.
>
>
>
> Thanking you in advance,
>
> Susana Kalko