Choosing right GSEA background gene sets

Question

[yxwucq]

I need to test whether a particular gene set is upregulated in my data. First, I have to provide a sorted ranked background gene set like this:

gene_name logfc
xx 3.0
yy 2.0

I can get this ranked list from different expression analysis, but do I need to filter out insignificant genes (based on p-value or logfc) in advance? Or just provide all detected genes?

[jv]

no filtering for GSEA other than filtering of low expressed genes for your differential expression analysis.

The following paper offers some additional insights: Urgent need for consistent standards in functional enrichment analysis

Answer 1

[ATpoint]

Run on all tested genes. Here is why: Can I run GSEA on a subset of genes