Hi, I have a gene expression data in cpm not raw count and I want to do DE analysis using edgeR. In this case, is it okay to follow the overall workflow shown in the edgeR user guide because it's not raw counts in integer...? I'm wondering if it's okay to do "calcNormFactors(y)" where y is the DGEList of the data in cpm.

x is my data that contains gene expression values in cpm.

x <- read.table (file~) : this is cpm values of each gene in each samples.`
y <- DGEList(x)
y <- calcNormFactors(y)

thanks in advance.

No, it’s not. If you can’t get the raw counts, then the only thing you can really do is to add a small prior count, log2 transform the data, then use the limma- trend pipeline, ie. use eBayes(fit, trend=TRUE)

You can find more details in the limma help pages, as well as searching these forums.

It is, in fact, so wrong that we have a specific section of the edgeR user's guide (2.7.6) warning against it.