> Date: Fri, 2 Dec 2005 16:09:13 +0100 (CET) > From: kfbargad at ehu.es > Subject: [BioC] three group ANOVA > To: Bioconductor <bioconductor at="" stat.math.ethz.ch=""> > > > Dear List > > I am trying to find out if the mean expression values of Q-PCR differ > for a group of 192 genes in three different sample groups. I am mainly > interested in the F value, not any particular contrasts. I have > created an exprSet out of my QPCR > data using the following > >>myEset > > Expression Set (exprSet) with > 192 genes > 40 samples > phenoData object with 1 variables and 40 cases > varLabels > cov1: read from file > > > How can I proceed now? I thought of using linear models and then > obtain the F value from fit3$F.p.value for the fit, but I got the > following errors: > >>treatments = factor(c > (0,1,1,1,1,1,0,0,1,1,0,0,2,0,0,1,1,0,2,2,1,0,0,0,2,2,0,0,1,0,1,0,0,1 ,0, > 0,0,0,0,0),labels = c("N0","N1","N2")) >>design = model.matrix(~0+treatments) >>fit <- lmFit(exprs(myEset),design) >>contrast.matrix = makeContrasts(N1-N0,levels = design) >>fit2 = contrasts.fit(fit,contrast.matrix) >>fit3 = eBayes(fit2) > Error in ebayes(fit = fit, proportion = proportion, stdev.coef.lim = > stdev.coef.lim) : > No residual degrees of freedom in linear model fits > > > Best, > > David Dear David, A good way to proceed would be to look at the quantities you have created to check they are as you expect. Try printing out your design matrix. Does it have columns called 'N0' or 'N1'? If not ... Best wishes Gordon