EdgeR confusion with no replicates

[Studentlost]

I have four samples(no replicates) in a merged raw counts file from HTSeq. I'm confused on how to create the DGEList and counts via rnbinomial in order to analyze expression levels. The vignette wasn't too clear and I'm not very fluent in R or statistics.

After reading the raw counts file in and assigning groups I'm not sure what is supposed to be entered for counts.

Do you guys have a script you can share or can you walk me through the first few lines just to get the object loaded so that I can analyze the data?

Thank you so much for your help!

[sdriscoll]

Hey there. I usually go about using edgeR a little differently. You can go in with a counts matrix (or data frame) and a single vector defining the conditions of your samples then perform the entire edgeR process like this:

d <- DGEList(counts=counts_table, group=condition_factors)
d <- calcNormFactors(d)
d <- estimateCommonDisp(d)
d <- estimateTagwiseDisp(d)
d.res <- exactTest(d, pair=c("A", "B")) # where A and B are conditions

That being said, edgeR does not like unreplicated experiments. I believe it requires that at least one condition is replicated at minimum.