GSEA to discover mixed enriched terms

[Maurits Unkel]

Hi,

Now, enriched terms are either up- or down regulated, likely because of the weighting in the running sum implementation. However, there must be terms with mixed enrichment, where some type of mixed activation/inhibition actually leads to very enriched terms, which is now missed through the implementation. I realise this is more in the domain of topology based algorithms, applied to terms which represent pathways, however, I'm hopeful we could get some additional information out of just the ranked expression list.

In order to find mixed enriched terms, I'm curious if it would be proper, and statistically viable, to make the expression values absolute before running GSEA? 

Or would additional transformations like scaling/normalization be needed additionally? 

Best,

Maus

[Anthony Castanza]

Hi Maus,

Yes, the inability to detect perturbations in gene sets where there are distinct components that are regulated in opposite directions is a known limitation of the standard methods of GSEA. The GSEA desktop application does support an absolute value mode to help address this, it is under "Gene list sorting mode" in the Basic fields section. It computes the ranking metrics as you would normally, but then performs enrichment calculations on the basis of the absolute value of the ranking metric. I can't really vouch for it personally, but in theory it is a reasonable thing to do. That mode isn't supported for GSEA Preranked however.

Unfortunately I also can't really speak to additional normalizations, you might explore something like a z-score transform if you were to construct your own absolute value ranking for GSEA-preranked, but this isn't something we necessarily advise.

-Anthony

Anthony S. Castanza, PhD

Curator, Molecular Signatures Database

Mesirov Lab, Department of Medicine

University of California, San Diego

--
You received this message because you are subscribed to the Google Groups "gsea-help" group.
To unsubscribe from this group and stop receiving emails from it, send an email to gsea-help+...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/gsea-help/6376530c-764a-460d-bc78-fec90e099101n%40googlegroups.com.

[Maurits Unkel]

Hi Anthony,

Where could I find the implementation code of the "Gene list sorting mode"? Or is it simply doing abs(expression values) and basically nothing else? 

I'll take a look at z-score transform! 

Thanks,

Maus

Op vrijdag 17 mei 2024 om 22:18:01 UTC+2 schreef Anthony Castanza:

[Anthony Castanza]

Hi Maus,

I believe that the code for these specific functions is located here:
https://github.com/GSEA-MSigDB/gsea-desktop/blob/8319c1e94cbefa880f3e62127ef8b813d1ff273f/src/main/java/edu/mit/broad/genome/math/Vector.java

note that the GSEA desktop application is in Java.

Based on a cursory review of these functions it does, in fact, appear to largely be a simple absolute performed after the basic ranking procedure.

-Anthony

Anthony S. Castanza, PhD

Curator, Molecular Signatures Database

Mesirov Lab, Department of Medicine

University of California, San Diego

To view this discussion on the web visit https://groups.google.com/d/msgid/gsea-help/fce84217-ed5b-4400-8daf-0a25c078374bn%40googlegroups.com.