GSEA analysis RNA-seq data

Question

[Marco Pannone]

Hello!

I am performing the GO/pathway analysis of my RNA-seq data using GSEA. I have many DEGs, about 3000 in total, and I got many enriched pathways with both positive and negative NES. However, all or most of the enriched pathways have an FDR > 0.05 and therefore are not classified as statistically significant. Most of such enriched pathways are characterized by more than 30 DEGs, which is a reasonable number of differentially expressed genes associated with a certain pathway.

What could be the reason for that?

I would gladly appreciate any comment or suggestion and thanks in advance!

Answer 1

[MatthewP]

How did you perform GSEA? What's your input data? You need to input ALL genes data to GSEA not only DEGs.

[Marco Pannone]

I am running GSEA through the WebGestalt website, where it asks me to report a list of two columns: one should contain the gene identifier and another one the associated "scores", which I assume are the log2FC values of the retrieved DEGs. I do upload together both the UP and DOWN DEGs.

Is there something wrong I am doing?

[MatthewP]

Log2FC is fine. You need to upload all genes, including those not DEGs.

[Marco Pannone]

Thanks a lot! I will do so then. I was including only the DEGs with a log2FC threshold of > 0.5 for UP DEGs and < -0.5 for DOWN DEGs. Including all the genes means that I also have to remove the adjusted pvalue threshold? I have tried this way and some pathways now pass the FDR threshold and are classified as statistically significant.