Dear list, I am comparing cells with a gene knockdown with wildtype cells to find genes differentially regulated in response to the knockdown using Illumina bead arrays. I plan to compare genes regulated by CPT stimulation in knockdown and wildtype cells. For that, I used a Venn Diagram, and export a list of the genes of the knockdown only category. However, whatever I enter for lfc in the decideTests function, the numbers of genes in the resulting diagram stay the same, whilst from manual comparison of the data, they should not. Does anyone know what I am doing wrong? Many thanks, Moritz (Fellow, University of Bonn, Germany) I did the following, largely following the beadarray tutorial. library(beadarray) dataFile = "data.txt" BSData = readBeadSummaryData(dataFile = dataFile, dec=",", qcFile = NULL, skip = 0, qc.skip = 0) BSData.quantile = normaliseIllumina(BSData, method = "quantile", transform = "log2") BSData.genes = BSData.quantile[which(fData(BSData)$Status == "Gene"), ] expressed = apply(Detection(BSData.genes) < 0.05, 1, any) BSData.filt = BSData.genes[expressed,] library(limma) samples = c(rep("con_veh",3), rep("con_cpt",3), rep("kd_veh",3), rep("kd_cpt",3) ) samples samples = as.factor(samples) design=model.matrix(~0 + samples) colnames(design) = levels(samples) fit = lmFit(exprs(BSData.filt), design) cont.matrix=makeContrasts(diff_con=con_veh - con_cpt, diff_kd=kd_veh - kd_cpt, levels=design) fit=contrasts.fit(fit, cont.matrix) fit$genes=fData(BSData.filt) ebFit=eBayes(fit) veh=topTable(ebFit, coef=1, number=500) cpt=topTable(ebFit, coef=2, number=500) write.table(veh, file="diff_exp_wt.txt", sep="\t") write.table(cpt, file="diff_exp_kd.txt", sep="\t") decideTests(ebFit,method="separate",adjust.method="BH",p.value=0.05,lf c=1) results=decideTests(ebFit) a <- vennCounts(results) print(a) vennDiagram(a) only_kd <- which(results[,1] == 0 & results[,2] == 1) [[alternative HTML version deleted]]

On Tue, 9 Nov 2010 15:24:48 +0100, Moritz Kebschull <endothel at="" gmail.com=""> wrote: > Dear list, > > I am comparing cells with a gene knockdown with wildtype cells to find > genes > differentially regulated in response to the knockdown using Illumina bead > arrays. I plan to compare genes regulated by CPT stimulation in knockdown > and wildtype cells. For that, I used a Venn Diagram, and export a list of > the genes of the knockdown only category. > However, whatever I enter for lfc in the decideTests function, the numbers > of genes in the resulting diagram stay the same, whilst from manual > comparison of the data, they should not. Does anyone know what I am doing > wrong? > [snip] > decideTests(ebFit,method="separate",adjust.method="BH",p.value=0.05,lf c=1) > results=decideTests(ebFit) > a <- vennCounts(results) Look closely at these lines! -- Alex Gutteridge

On Tue, 9 Nov 2010 15:24:48 +0100, Moritz Kebschull <endothel at="" gmail.com=""> wrote: > Dear list, > > I am comparing cells with a gene knockdown with wildtype cells to find > genes > differentially regulated in response to the knockdown using Illumina bead > arrays. I plan to compare genes regulated by CPT stimulation in knockdown > and wildtype cells. For that, I used a Venn Diagram, and export a list of > the genes of the knockdown only category. > However, whatever I enter for lfc in the decideTests function, the numbers > of genes in the resulting diagram stay the same, whilst from manual > comparison of the data, they should not. Does anyone know what I am doing > wrong? > [snip] > decideTests(ebFit,method="separate",adjust.method="BH",p.value=0.05,lf c=1) > results=decideTests(ebFit) > a <- vennCounts(results) Sorry, if my previous post was too cryptic. I know if you stare at these things long enough you miss the wood for the trees. You need to assign the result of your first call to decideTests() to uh... "results". And then remove the second call to decideTests(). E.g. results <- decideTests(ebFit,method="separate",adjust.method="BH",p.value=0.05,lf c=1) a <- vennCounts(results) Should get you where you want to be. -- Alex Gutteridge