GSEA with GO and KEGG datasets

Question

[shahbaz.ahmed]

Hello everyone,

I'm trying to understand how GSEA works and what I have understood so far is that in GSEA we use all the genes in our RNA-seq data set and then rank them with the most expressed at the top and underexpressed at the bottom and then find the genes for particular function lying in upregulated or downregulated group. it is different than GO and KEGG where we only use DEGs that are enriched for GO terms and KEGG pathways. I came across this statement in one of the analyses where they used GO and KEGG datasets for GSEA, generating two GSEA results folders, each for GO and KEGG. this confuses me since I haven't heard in any of the tutorial people mentioning if we use the GO and KEGG datasets too for GSEA, I thought we use all the genes in our dataset. can someone please help me understand the concept?

Thank you

Answer 1

[Papyrus]

Gene set/pathway enrichment analyses are used to interrogate gene sets (groups of genes) in your data. You have to differentiate two things:

• There are many types of gene/pathway enrichment analyses (see this review for example). Two of the most used analyses are, for example, ORA methods (over representation analyses) and GSEA (functional class scoring).
• On the other hand, all these analyses use databases describing pathways, such as GO or KEGG.

In ORA, you typically use your set of DEGs and interrogate pathways (such as GO pathways, or KEGG pathways). In GSEA, you use your set of ALL genes (ranked) and interrogate pathways (again, such as GO pathways, or KEGG pathways).

I think the confusion is that, when people say the did a "GO enrichment analysis", they often actually want to say that they did a "ORA (analysis) using the GO pathway database", while when they say that "we did a GSEA", you always know which method they used (a functional class scoring method).

[shahbaz.ahmed]

Thank you so much!! this clears my confusion