I have a question about the GSVA package, In particular about the results with different methods parameter value ("ssGSEA" and "GSVA" ). In my case I have 5 sample and I want to test the enrichment of 2 genesets on them. The step of my analysis was the following:
- produce a raw counts matrix for BAM files
- normalise the raw counts (DESeq2)
- transform the normalised counts via regularised log
- Run GSVA
I've performed the gsva() 2 times with same parameter except for the method; one time i used ssGSEA and for the other GSVA, the results i've obtained led to opposite conclusions (regarding the enrichment of the two genesets). I've attached the plot showing the obtained scores
On the left GSVA score plot and ssGSEA on the right.
My question is, it's possible that the two methods gives me such opposite results when described as "similar" in the help page? I've read the reference paper trying to understand better which could be the source of the variation but i think i'm loosing something (or more), it's probably due to the gene ranking step? even if it's due to that or from other aspects of the computation i didn't expect, from what i know, such a difference in results. If someone can fill my gap in knowledge for big it is and give me some insights/explanations on the argument and results understanding it would be awesome. Let me know if more information are needed from me on the problem.