GSEA pathway analysis results on the v6.0 are different from those I got previously on the v5.0, whereas results are the same as before using Mummichog.
Does it make sense? Was the GSEA workflow modified but not that of Mummichog?
Thanks for your help,
You need to provide details. Keep in mind that all the underlying pathways are updated.
I re-ran a few pathway analyses on LC-MS data (peak intensity tables) using the “Functional Analysis” tool of MetaboAnalyst v6: I used the same data matrices and selected the same parameters (missing values, data filter, normalization, etc.) as I had done last year (feb 2023) on the v5 version.
Today, when selecting the Mummichog algorithm on the “set parameter” page, I get the exact same list of pathways with the same number of hits, same scores and p-values.
When I select the GSEA algorithm I get the same number of pathways (and hits) as before, yet with much better p-values and NES scores. This is actually a rather good news, but I’d like to be sure before re-running other analyses.
I’ve tested all possible parameter combinations I could think of to double-check I hadn’t done something differently, but the fact that Mummichog gives the same results as before is very confusing. I also find it weird that the update of pathways would only impact the p-values of GSEA.
I hope this makes sense, thanks for your help!
NB: the export button to generate the “pathway hits” file doesn’t seem to be working at the moment (although the one for “compound hits” work).