Hi everyone,
I have a question and I was hoping to get some insight from you.
I ran RNA-seq on my samples and I didn't have replicates, I used differentially expressed genes with a cut-off of 2-fold change for Preranked-GSEA and got a list of pathways activated in each of my samples. The question is that can I use any of the values such as ES, NES, FDR, etc. or since I have no replicates, it doesn't make sense to use these? Next, if I don't use these values, should I rank my pathways based on the number of genes they have in the gene set? If yes, is there a cut-off that is being commonly used for that?
Thanks for the help.
As long as I have understood you have only one replicate per sample group, is it right? If so, how did you compute significant DEGs? With what statistical test?
In your case, if you are interested in highlighting to which pathways your genes of interest contribute the most to, you can better perform an Over-Representation Analysis (ORA), rather than running any pre-ranked or full GSEA.
An easy way to perform ORA is on http://www.webgestalt.org.
Without replicates I would not really make any significant conclusions out of the differential analysis. I would say you minimum need 3 high quality replicates, in order to perform a robust analysis.
I would say that at least three replicates per sample are needed. Otherwise, you can rank the pathways but can't show statistical significance. But I feel manual ranking will be tedious.
- Badges
- Science topic
- Similar topics
- Filing