Dear Members, I have this dataset extracted from GEO database (GSE16962). According to the data information provided below, I designed a matrix for lmFit. I am not sure if it is correct. Could somebody take a quick look? > show(pData(phenoData(gse[[1]]))[1:12,c(1,6,8)]) title type source_name_ch1 GSM424759 pSUPER-scramble replicate 1 RNA HUVEC infected by retroviral vectors bearing a control scramble sequence GSM424760 pSUPER-scramble replicate 2 RNA HUVEC infected by retroviral vectors bearing a control scramble sequence GSM424761 pSUPER-scramble replicate 3 RNA HUVEC infected by retroviral vectors bearing a control scramble sequence GSM424762 pSUPER-mir-210 replicate 1 RNA HUVEC infected by retroviral vectors bearing pre-miR-210 sequence GSM424763 pSUPER-mir-210 replicate 2 RNA HUVEC infected by retroviral vectors bearing pre-miR-210 sequence GSM424764 pSUPER-mir-210 replicate 3 RNA HUVEC infected by retroviral vectors bearing pre-miR-210 sequence GSM424765 Anti-scramble replicate 1 RNA HUVEC transfected with a scramble-LNA GSM424766 Anti-scramble replicate 2 RNA HUVEC transfected with a scramble-LNA GSM424767 Anti-scramble replicate 3 RNA HUVEC transfected with a scramble-LNA GSM424768 Anti-mir-210 replicate 1 RNA HUVEC transfected with anti-miR-210-LNA GSM424769 Anti-mir-210 replicate 2 RNA HUVEC transfected with anti-miR-210-LNA GSM424770 Anti-mir-210 replicate 3 RNA HUVEC transfected with anti-miR-210-LNA >treatments=factor(c(1,1,1,2,2,2,3,3,3,4,4,4), labels=c("CTRL","mir210","CTRL1","CTRL2")) > design=model.matrix(~treatments) > colnames(design)=c("CTRL","mir210","CTRL1","CTRL2") > design CTRL mir210 CTRL1 CTRL2 1 1 0 0 0 2 1 0 0 0 3 1 0 0 0 4 1 1 0 0 5 1 1 0 0 6 1 1 0 0 7 1 0 1 0 8 1 0 1 0 9 1 0 1 0 10 1 0 0 1 11 1 0 0 1 12 1 0 0 1 >library(limma) > fit=lmFit(eset,design) > fit=eBayes(fit) Warning message: Zero sample variances detected, have been offset > results=classifyTestsF(fit,p.value=0.0001) > summary(results) CTRL mir210 CTRL1 CTRL2 -1 0 1364 733 961 0 0 45192 48699 48787 1 54675 8119 5243 4927 ###mir210 and CTRL2 are the two arrays that I really care about. CTRL and CTRL1 are experiment control. CTRL2 may have some impact to true result. So I decided to make following cont.contrasts >cont.contrasts=makeContrasts(mir210=c(0,1,0,0),CTRL2=c(0,0,0,1),level s=design) > cont.contrasts Contrasts Levels mir210 CTRL2 CTRL 0 0 mir210 1 0 CTRL1 0 0 CTRL2 0 1 > fit2=contrasts.fit(fit,cont.contrasts) > fit2=eBayes(fit2) > results=classifyTestsF(fit2,p.value=0.0001) > summary(results) mir210 CTRL2 -1 331 559 0 54151 53908 It looks to me that this is the TRUE differentially expressed list for mir210 treatments. 1 193 208 > table(Cmir210=results[,1],CTRL2=results[,2]) CTRL2 Cmir210 -1 0 1 -1 17 272 42 0 450 53535 166 1 92 101 0 Many Many thanks Jing [[alternative HTML version deleted]]