[BioC] automating makeContrasts call in limma

[Gordon Smyth]

You need

   makeContrasts(contrasts=con, levels=dMatrix)

See the help page.

Gordon

>Date: Thu, 25 Oct 2007 12:46:57 +0200
>From: Amin Moghaddasi <a.moghaddasi at gmail.com>
>Subject: [BioC] automating makeContrasts call in limma
>To: bioconductor at stat.math.ethz.ch
>
>Hi list,
>
>I am trying to call makeContrasts() in limma to accept character vector
>of all samples vs control. But it seems that makeContrasts() consider
>only the first element in the contrast vector.
>
>Basically what I've done is first to assign design matrix "dMatrix",
>then make a contrast character vector "con" using paste() like:
>
> > dMatrix <- model.matrix(~ -1+factor(c(1,1,2,2,3,3)))
> > colnames(dMatrix) <-  paste("s", 1:(dim(dMatrix)[2]), sep = "")
> > con<-paste(rep('s'),sep="",3:1,"-s1")
> > con
>[1] "s3-s1" "s2-s1" "s1-s1"
>
>Then call makeContrasts like:
> > cMatrix <- makeContrasts(con, levels=dMatrix)
> > cMatrix
>       Contrasts
>Levels s3-s1
>     s1     0
>     s2     0
>     s3     0
>
>The expected result is :
> > cMatrix <- makeContrasts("s3-s1","s2-s1","s1-s1", levels=dMatrix)
> > cMatrix
>       Contrasts
>Levels s3-s1 s2-s1 s1-s1
>     s1    -1    -1     0
>     s2     0     1     0
>     s3     1     0     0
>
>So any advise of what I'm doing wrong would be appreciated, Or how can I
>automatically call makeContrasts using a vector of all samples vs
>control.
>
>Thanks in advance,
>Amin.