edgeR for paired cases

[lran2008]

Hi all,

I am using EdgeR for DE analysis. I would appreciate if somebody could help me check my experiment design.

I have 6 cows (2,63,66,67,73,74) for a treatment, each with 3 time points(m14,p7,p28). I want to compare between p28 vs m14, p7 vs m14 and p28 vs p7.


cow <- factor(c(2,2,2,63,63,63,66,66,66,67,67,67,73,73,73,74,74,74))
timep <- factor(c("m14","p7","p28","m14","p7","p28","m14","p7","p28","m14","p7","p28","m14","p7","p28","m14","p7","p28"))
data.frame(Sample=colnames(y),cow,timep)
design<-model.matrix(~cow+timep)

This follows the EdgerR user guide (31 March 2013 revised), case study 4.4 on page 53.

Thanks.

[dpryan]

The short answer is that you want to simply use constrasts, such as accepted by glmLRT. Here's one thread on the bioconductor email list (the edgeR authors are pretty active there) that gives some suggestions to get you started.