Gene set enrichment analysis

Question

[Bogdan]

Dear all,

we wish you a fruitful autumn time ! I would like to ask for a piece of advice, particularly, which package would you recommend for GSEA (preferably in R/BioC, and it is close to java implementation available at http://software.broadinstitute.org/gsea/doc/desktop_tutorial.jsp ).

many thanks for your suggestions,

-- bogdan

[Ram]

Have you tried WebGestalt or its R implementation, WebGestaltR? We have a new version with a much more intuitive UI!

[Bogdan]

Thank you very much for your suggestions.

I have been looking into WebGestaltR documentation (https://cran.r-project.org/web/packages/WebGestaltR/WebGestaltR.pdf), and if I may ask, what is the simplest setup of the command "WebGestaltR" ?

In the package manual, it is written :

WebGestaltR(enrichMethod="ORA", organism="hsapiens",
    enrichDatabase="geneontology_Biological_Process",enrichDatabaseFile=NULL,
    enrichDatabaseType=NULL,enrichDatabaseDescriptionFile=NULL,interestGeneFile=NULL,
    interestGene=NULL,interestGeneType=NULL,collapseMethod="mean",referenceGeneFile=NULL,
    referenceGene=NULL,referenceGeneType=NULL,referenceSet=NULL, minNum=10, maxNum=500,
    fdrMethod="BH",sigMethod="fdr",fdrThr=0.05,topThr=10,dNum=20,perNum=1000,
    lNum=20,is.output=TRUE,outputDirectory=getwd(),projectName=NULL,keepGSEAFolder=FALSE,
    methodType="R",dagColor="binary",hostName="http://www.webgestalt.org/")

[Ram]

All the parameters with recommended values have them as default values. You can check out the Webgestalt user forum here. I'm sorry, I'm not too familiar with WebgestaltR.

[Bogdan]

Thanks a lot, Ram !

yes, we would like to include Webgestalt (preferably WebgestaltR) in our R computational pipeline.

[Trypto]

Hi Bogdan,

We just recently implemented GSEA and it is still in development branch. You can check it out at here. It is designed to add weights to each "gene" in the "gene set", but can also perform standard GSEA as well. I just added a utility function to help load and format the input. It is pure R now and feels like a little bit slower than original.

If you just need to run GSEA with your own data, I think you can take the GSEA file alone for now. We are going to merge dev into master branch and release a new version on CRAN later this year.

(The beta version of WebGestalt already uses the new version in the backend. One of the benefit is that we can get leading edge genes and make GSEA enrichment plot for all the gene sets, not limited by -plot_top_x parameter, for example.)

Answer 1

[Malcolm.Cook]

I've used both Broad's JAVA application and clusterProfiler's R implementation of GSEA and recommend it as performing the same analysis and producing near identical charts. Also, if you adopt it, there are a a bunch of other analyses and visualizations which become available with single setup (e.g. GO & Kegg, in addition to GSEA). I find the documentation a little terse, but everything is there you will need.

[Bogdan]

Thanks, Malcolm ! Just came across EnrichmentBrowser, that seems to have implemented lots of algorithms :

https://bioconductor.org/packages/release/bioc/html/EnrichmentBrowser.html