Standard GSEA and RNA-seq data

Question

[pennakiza]

Hi all,

I am a bit confused with GSEA and I was hoping that you could help me. I've got RNA-Seq data (two conditions with 5 replicates each). The GSEA website says that the suggested method is the standard one vs the pre-ranked so I decided to give it a try. I have read the following on this paper: "Gene set permutation is recommended for studies with limited variability and biological replicates (i.e., two to five per condition). In this case, differential gene expression values should be computed outside of GSEA, using methods that include variance stabilization (such as edgeR29, DESeq30 and limma/voom31,32) and imported into the GSEA software before pathway analysis."

I'm not sure, should I do a standard GSEA with phenotype or a gene set permutation?

Thanks!!

-P

Answer 1

[Kevin Blighe]

Yes, derive the differential expression test statistics outside of GSEA, assuming that you are referring to Broad Institute's program called 'GSEA', and not the general technique of GSEA (gene-set enrichment analysis).

I would use DESeq2 with the variance-stabiling transformation.

Kevin

[pennakiza]

Thanks Kevin,

Yes, I am referring to the program "GSEA" by the Broad Institute. When you say you would use the DESeq2 vst you mean the table of transformed counts to do a standard GSEA, instead of the Pre-Ranked GSEA?

Sorry for my confusion!

-Penny

[Kevin Blighe]

Hey Penny, I had assumed that the idea was to perform the statistical comparisons using DESeq2 for the purpose of pre-ranking genes, and to also use the VST (variance stabilising transformation) expression values as input.

I have used GSVA more than Broad's GSEA ,though.

[pennakiza]

I might be completely wrong here but if I understand it correctly for the pre-ranked test one just loads a list of genes ranked by the shrunken log2 fold change values, along with those values. Is that wrong?

Excuse my ignorance please!

Thanks,

Penny

[Kevin Blighe]

Indeed, there are different ways of performing GSEA. You know better than I. Please do not feel that you are ignorant about using the program.