Dear Irazoki, You have a warning from eBayes that suggests a major problem with your data, independently of what contrasts you are taking. Time to do some basic quality assessment of your data. You need to examine some sensible plots of your data before rushing into hypothesis testing. Best wishes Gordon > Date: Tue, 21 May 2013 07:48:13 -0700 (PDT) > From: "Oihane Irazoki [guest]" <guest at="" bioconductor.org=""> > To: bioconductor at r-project.org, o.irazoki at gmail.com > Subject: [BioC] problem with contrast.matrix and eBayes in limma > > > Hi! I'm sorry for bothering you. I'm a new R-user and I'm having some > problems while doing a microarray analysis. I'm comparing the whole > genome array of a Salmonella serovar to another 25, and my goal is to > determine which genes are differentially expressed. I'm using limma > package and running the next code, > > > DIFFERENTIAL EXPRESSION i > #several groups comparisson (by serovars) > > groups <- read.table("ClusteringSamples.txt", head=T, sep='\t') > > f <- factor(groups$Serovar) > design <- model.matrix(~0 + f) > > colnames(design) <- c("Abortusovis", "Agona", "Anatum", "Arizonae", "Braenderup", > "Bredeney", "Cholerasuis", "Derby", "Enteritidis", "Gallinarum", "Goelzau", > "Hadar", "Havana", "Infantis", "Kedougou", "Mbandaka", "Mikawasima", "Ohio", > "ParatiphyA", "ParatiphyB", "Pos.Control", "Pullorum", "Typhi", > "Typhimurium", "Virchow") > #Convertir a vectores! > > summaryis.na(data)) #check if there is any missing value in the dataset. > > fit <- lmFit(data, design) > > contrast.matrix <- makeContrasts(Abortusovis-Pos.Control, Agona- Pos.Control, > Anatum-Pos.Control, Arizonae-Pos.Control, Braenderup- Pos.Control, > Bredeney-Pos.Control, Cholerasuis-Pos.Control, Derby- Pos.Control, > Enteritidis-Pos.Control, Gallinarum-Pos.Control, Goelzau- Pos.Control, > Hadar-Pos.Control, Havana-Pos.Control, Infantis-Pos.Control, > Kedougou-Pos.Control, Mbandaka-Pos.Control, Mikawasima- Pos.Control, > Ohio-Pos.Control, ParatiphyA-Pos.Control, ParatiphyB- Pos.Control, > Pos.Control-Pos.Control, Pullorum-Pos.Control, Typhi- Pos.Control, > Typhimurium-Pos.Control, Virchow-Pos.Control, > levels=design) > > fit1 <- contrasts.fit(fit, contrast.matrix) > fit2 <- eBayes(fit1) > > But when I try to run eBayes correction to then compute topTable and get those differentially expressed genes, I get the next error back: > > > Error in eigen(cor.matrix, symmetric = TRUE) : > infinite or missing values in 'x' > In addition: Warning messages: > 1: In ebayes(fit = fit, proportion = proportion, stdev.coef.lim = stdev.coef.lim, : > Estimation of var.prior failed - set to default value > 2: In cov2cor(object$cov.coefficients) : > diag(.) had 0 or NA entries; non-finite result is doubtful > > I try to eliminate those missing values, but then I only can compute the contrast matrix of 13 of the 25 different serovars. > How can I solve the problem? I'll appreciate all the help and advices. > > Oihane > > -- output of sessionInfo(): > > R version 3.0.0 (2013-04-03) > Platform: x86_64-apple-darwin10.8.0 (64-bit) > > locale: > [1] en_US.UTF-8/en_US.UTF-8/en_US.UTF-8/C/en_US.UTF-8/en_US.UTF-8 > > attached base packages: > [1] splines parallel grid stats graphics grDevices utils > [8] datasets methods base > > other attached packages: > [1] venneuler_1.1-0 rJava_0.9-4 siggenes_1.34.0 multtest_2.16.0 > [5] Biobase_2.20.0 BiocGenerics_0.6.0 limma_3.16.1 gplots_2.11.0 > [9] MASS_7.3-26 KernSmooth_2.23-10 caTools_1.14 gdata_2.12.0 > [13] gtools_2.7.1 ggplot2_0.9.3.1 > > loaded via a namespace (and not attached): > [1] bitops_1.0-5 colorspace_1.2-1 dichromat_2.0-0 digest_0.6.3 > [5] gtable_0.1.2 labeling_0.1 munsell_0.4 plyr_1.8 > [9] proto_0.3-10 RColorBrewer_1.0-5 reshape2_1.2.2 scales_0.2.3 > [13] stats4_3.0.0 stringr_0.6.2 survival_2.37-4 tools_3.0.0 > ______________________________________________________________________ The information in this email is confidential and intend...{{dropped:4}}