clusterProfiler::gseGO order of genes

[christophista]

Hello

I am using clusterProfiler (version for GSE analysis with the function clusterProfiler::gseGO. The parameter geneList is a list of log2FC sorted (lowest to highest, e.g., the down-regulated genes first, the up-regulated genes last) with the names corresponding to the gene symbols.

When the ranks are inverted (i.e., the log2FC is calculated not condition / control, but control / condition), the results of the GSEA change slightly, 6 terms of 352 are no longer significantly enriched.

My question is:
Is it expected the results (pure number of sig. enriched terms) change (marginally) when the test is done the other way round?
I would not expect that, but I assume it has to do with the random walk.

Please let me know your ideas about this or if you have any questions.

Thank you in advance!