**10**wrote:

Hello everyone

For a group project, we had to process a raw micro array data set using bioconductor and draw some conclusions. We chose this data set for our project. We looked a bit at the data and preprocessed it, but when we wanted to test for differential expressed genes, we got some strange results. Only a couple of genes showed up as DE, where we expected a lot more. The last few days we've searched through our code for errors and possible explanations, but we don't seem to find any differences in method of working from the examples we got. Can anyone could spot an obvious error in our code or give us another way to work? The code shows the part where we tested for DE genes.

#Obtain expression values ########################## biocLite("affyPLM") library("affyPLM") eset<-threestep(affy.data,background.method="RMA.2",normalize.method="quantile", summary.method="median.polish") #obtain the expression values on gene level with the function exprs() e<-exprs(eset) head(e) dim(e) #construction of design matrix library(limma) treatment<-factor(c(rep("Tis11withLPS",3),rep("Tis11",3),rep("GFPwithLPS",3),rep("GFP",3))) X<-model.matrix(~0+treatment) colnames(X)<-c("GFP","GFPwithLPS","Tis11","Tis11withLPS") #fit linear model lm.fit<-lmFit(e,X) #construction of contrasts mc<-makeContrasts("Tis11withLPS-GFPwithLPS","Tis11-GFP","Tis11withLPS-Tis11","GFPwithLPS-GFP", levels=X) c.fit<-contrasts.fit(lm.fit,mc) eb<-eBayes(c.fit) toptable(eb, sort.by = "logFC") topTableF(eb) #Extract p-values & adjustment modFpvalue <- eb$F.p.value indx <- p.adjust(modFpvalue, method = "bonferroni") < 0.05 sig <- modFpvalue[indx] nsiggenes <- length(sig)

**17k**• written 6 months ago by Lennart.Vermander •

**10**