[BioC] Limma: normalizeBetweenArray and Tquantile

[Gordon K Smyth]

> Date: Mon, 14 Apr 2008 18:39:21 +0800
> From: "Ng Stanley" <stanleyngkl at gmail.com>
> Subject: [BioC] Limma: normalizeBetweenArray and Tquantile
> To: bioconductor at stat.math.ethz.ch
>
> Hi,
>
> Mine arrays are dye-swapped. I would like to separate them into two groups
> and normalize the first group with respect to the green channel and the
> second by the red channel; the common reference is labelled green and red
> throughout the group.

This might not necessarily be a good idea.  Generally speaking, you should 
embark on this sort of custom normalisation only if you are an expert and 
you know what you're doing.

> I have two options:
>
> A) Use method="Gquantile" and "Rquantile", however, I am unsure how to
> combine the output of both together for later downstream analysis

Why not simply cbind the results back together?  E.g.,

   MA1 <- normalizeWithinArrays(MA[,group1],method="Gq")
   MA2 <- normalizeWithinArrays(MA[,group2],method="Rq")
   MA.norm <- cbind(MA1,MA2)
   targets.norm <- rbind(targets[group1,],targets[group2,])

> B) Use method="Tquantile". It needs a vector twice as long, but can't find
> any example how the vector should look like.

This doesn't do what you want.

Best wishes
Gordon

> Thanks
> Stanley