Hi I just wanted some advice on analyzing single channel data from Genepix in Limma. There are a number of slides that have bad cy5 signals and other chips where on cy3 was used, so I wanted to be able to just analyze the cy3 channel. After a search of the newsgroup, there was a post by Gordon that basically says use this: y2 <- normalizeBetweenArrays(RG$G, method="quantile") (or use vsn) and then run Limma normally. I was wondering if this is still the preferred method? If one just passes the cy3 channel values to lmFit, the weights don't seem to be passed along. Is there any way of combining the annotation, weights and expression values into an object that lmFit can recognize?

Dear Lance, Without creating a new object, you could pass the weights to lmFit() using the weights= argument, then pass the annotation to topTable() using the genelist= argument. If your annotation is just a single vector of gene IDs, you can add this to RG$G as rownames (it may complain if not unique). Alternatively, you could assemble an ExpressionSet object. Others can better advise you how to do that than me. Best wishes Gordon >Date: Fri, 11 May 2007 10:02:27 -0400 >From: "Lance E. Palmer" <lance.palmer at="" stonybrook.edu=""> >Subject: [BioC] Analyzing single-channel genepix data from Genepix in > Limma >To: bioconductor at stat.math.ethz.ch >Message-ID: <1178892147.26381.102.camel at informatics.bio.sunysb.edu> >Content-Type: text/plain > >Hi I just wanted some advice on analyzing single channel data from >Genepix in Limma. > >There are a number of slides that have bad cy5 signals and other chips >where on cy3 was used, so I wanted to be able to just analyze the cy3 >channel. > >After a search of the newsgroup, there was a post by Gordon that >basically says use this: > >y2 <- normalizeBetweenArrays(RG$G, method="quantile") >(or use vsn) >and then run Limma normally. > >I was wondering if this is still the preferred method? If one just >passes the cy3 channel values to lmFit, the weights don't seem to be >passed along. Is there any way of combining the annotation, weights >and expression values into an object that lmFit can recognize?