[BioC] LIMMA:linear model and contrast.fit

[Sebastian Thieme]

Hello !

I try to use LIMMA according to common reference design as descripted in 
the User's guide.
The data are stored in variable RGraw:

 >QAfile<-dir(pattern="*\\.txt")
 >RGraw<-read.maimages(QAfile, source="quantarray", wt.fun=wtIgnore.Filter)

The experiment has the following design:

 >targets <- readTargets("experimente.info")
 > targets
                 SlideNumber     Name    FileName        Cy3     Cy5
e78             1               e78     e78.txt         M1      G1
e73             2               e73     e73.txt         M1      G1
w21             3               w21     w21.txt         M1      G2
w18             4               w18     w18.txt         M1      G2
 >design <- modelMatrix(targets,ref="M1")
 >design
                 G1      G2
e78     1       0
e73     1       0
w21             0       1
w18     0       1

I normalize the following way:

 >MA<-normalizeWithinArray(RGraw,method="loess")

Then I want to make the linear modell like this

 >fit<-lmFit(MA,design=Design

but I get the error message:

Error in unwrapdups(M, ndups = ndups, spacing = spacing) :
         dim<- : dims [product 0] do not match the length of object [5760]

The following plan is:

 >contrast.matrix <- makeContrasts("G1","G2","G1-M1","G2-M1",levels=design)
 >fit2 <- contrasts.fit(fit, contrast.matrix)
 >fit2 <- eBayes(fit2)

I've no clue where the error is. Could anybody give a hint? And are the 
following steps useful?

Thanks!

Sebastian Thieme