[BioC] Limma B-statistics

[Gordon Smyth]

>Date: Mon, 25 Oct 2004 10:59:43 +0200
>From: "Ingunn Berget" <ingunn.berget at iha.nlh.no>
>Subject: [BioC] Limma B-statistics
>To: <bioconductor at stat.math.ethz.ch>
>Message-ID: <005a01c4ba70$f82e1690$9fb12780 at ihf4651>
>Content-Type: text/plain
>
>I have had the same problem as described below, but have not applied the 
>functions classifyTestsF() or decideTests()
>
>Here are my commands. They are almost exactly as in the the limma-tutorial
>
>festuca.norm0 is an object of class marrayNorm.
>The experiment is a dye.swop experiment with 2 arrays (one with each 
>labelling). There are two different types of samples on the array, and the 
>goal is to find the differential expressed genes.
>There are 5 replicateded spots of each gene on each array (556 genes in 
>total). Only spots with printed genes are included in the analysis.
>gene is a logical vector for if a spot is a gene or not.
>
>f.cor<-duplicateCorrelation(maM(festuca.norm0)[gene,],design=c(1,-1),ndups=5)
>fit <- 
>lmFit(festuca.norm0[gene,],design=c(1,-1),ndups=5,correlation=f.cor$cor)
>eb <- eBayes(fit)
>toptable(number = 25,genelist = gnames,fit = fit, eb = eb, adjust = "fdr")
>plot(fit$coef,eb$lods,xlab="Log2 Fold Change",ylab="Log Odds",pch=16,cex=0.2)
>I'm a beginner with R, Bioconductor (and microarrays), so I hope any 
>answers will give simple explanations/comments

You need to explain exactly why you think that there is a problem before we 
can help you. As far as we can tell, the commands you give here have worked 
correctly with no errors. If the problem is simply that you're not finding 
any significant differential expression, that is not in itself an 
indication of a software error!

Gordon

>------------------------------------------------------------------------------
>Ingunn Berget
>Agricultural University of Norway
>Department of Animal and Aquacultural Sciences