ssGSEA Score

[Carlos Biagi Jr]

Hi everyone,

Here in our lab we are using the Single Sample GSEA (ssGSEA) methodology using the R codes provided by GenePattern platform.

The tool allows us to work with a signature that contains upregulated and downregulated genes. Thus, analyzing the methodology in more detail, we note that when working with a signature divided into UP and DOWN genes, the UP genes score is subtracted from the DOWNs genes score, thus generating a single score for that signature .

I would like to know why performing the subtraction to join the score of the UPs and DOWNs genes? Why not use the sum, average, or some other math operation?

Do you advise some other way of working with the UPs and DOWNs genes? Separately or in some other way?

Thank you very much.

Bests,

Carlos Biagi Jr

Please respond with the following information as applicable:
 Complete description of the problem (ie what you were trying to do and what went wrong):

Job ID if run on a publicly available server.

If not on Public Server, please share your input files, if possible. If you need to do so privately, let us know.
 
   What were your Parameter values?
   Server Platform (Mac/Linux)?

   Any errors or warnings you see in either stderr.txt, stdout.txt or popup dialogues:


Please note that many common errors can be resolved by searching this forum or taking a look at our documentation: 

* Module Documentation - available from the "documentation" link in the upper right hand corner of the module input page or here: http://software.broadinstitute.org/cancer/software/genepattern/modules
* GenePattern FAQ - http://software.broadinstitute.org/cancer/software/genepattern/FAQ
* GenePattern  Videos - http://software.broadinstitute.org/cancer/software/genepattern/video-tutorials
* GenePattern File Formats - http://software.broadinstitute.org/cancer/software/genepattern/file-formats-guide#file-formats
* GenePattern User Guide - http://software.broadinstitute.org/cancer/software/genepattern/user-guide#user-guide
* GenePattern Programmer's Guide - http://software.broadinstitute.org/cancer/software/genepattern/programmers-guide
* GenePattern Release Notes - http://software.broadinstitute.org/cancer/software/genepattern/doc/relnotes/current

[David Eby]

Hi Carlos,

I reached out to the original author of the ssGSEA method.  Here is his answer:

The combination of UP and DN scores by subtracting the DN score from the UP score is the easiest way to reduce both to a single number. It assumes that the UP part will be up-regulated in the same cellular context as the DN part is down-regulated resulting in a larger enrichment score by combining them.  This doesn't always happens because of different reasons: asymmetry in the UP and DN phenotypes, different strength of the UP and DN regulation, lack of generalization of the DN signatures etc. For these reasons it makes sense sometimes to use them separately. 

I hope that's helpful!

Regards,

David

David Eby
Consultant
Cancer Informatics Development
Broad Institute of MIT and Harvard
415 Main St, Cambridge, MA 02142, USA
http://www.broadinstitute.org/cancer
http://www.gsea-msigdb.org
https://twitter.com/GSEA_MSigDB
https://twitter.com/GenePattern