Hi,

I have a question about interpreting contrast.

I have three conditions, for e.g. high, med and low. I am interesting in looking at the genes which are differentially expressed in high condition over others.

I am using following code to look at High vs Low

HvsL<-makeContrasts(high - low,levels=design)

qlf<-glmQLFTest(fit,contrast=HvsL)

topTags(qlf,n=15)

If there is a positive logFold change, does it mean they are up-regulated in high over low?

and if they are down-regulated (negative fold change) means highly expressed in low over high?

What's a rule of thumb? 

Thanks

There's nothing special going on here. The logFC returned by edgeR, limma, and any other linear modeling framework is simply the result of evaluating your contrast as an arithmetic expression, substituting the fitted coefficient values from your model. In your example, if the logFC is positive, this means that high - low > 0. Solving that inequality by adding low to both sides, we get high > low, meaning that the expression is higher in the "high" condition than the "low" condition.