[BioC] Design matrix: dye-swap or channel-swap?

Gordon Smyth smyth at wehi.edu.au
Fri Dec 12 00:33:46 MET 2003


At 10:23 AM 12/12/2003, John Welsh wrote:
>I am confused about the design matrix for gls.series in limma.
>I have four two-color microarrays comparing samples A and B.
>A is always in channel1 and B is always in channel2.
>Red is in the first channel in two of them, and in the other channel in the
>other two.

Your question is a bit confusing because, normally, one channel belongs to 
one dye and the other channel belongs to the other dye. Normally one 
doesn't swap dyes but not channels. The question is, when your data is read 
into limma, does 'A' always get read into the G matrix and 'B' always into 
the 'R' matrix? If so you want design matrix all 1's. If not, you need 1 
and -1. I am guessing that the former is true, but you haven't given enough 
information to tell for sure.


>Do I want the design matrix c(-1,1,-1,1) or c(1,1,1,1)?
>John Welsh
>Associate Professor
>Sidney Kimmel Cancer Center
>10835 Altman Row
>San Diego, CA 92121
>(858) 450-5990 ex.282
>jwelsh at skcc.org
>Bioconductor mailing list
>Bioconductor at stat.math.ethz.ch

More information about the Bioconductor mailing list