Hi Iain, Yes, that's right. The MSigDB gene sets are defined in terms of gene symbols, so you have no alternative but to annotate your features by gene symbol in order to define set membership. Di Wu has given more detailed commentary of the steps in a separate email. Best wishes Gordon > Date: Tue, 9 Nov 2010 13:39:09 +0000 (GMT) > From: Iain Gallagher <iaingallagher at="" btopenworld.com=""> > To: bioconductor <bioconductor at="" stat.math.ethz.ch=""> > Subject: [BioC] romer in limma question > > Hello list > > I'm being a bit stupid with romer (for GSEA) in the limma suite. In > order to run the analysis I need to indicate in each gene set I have > which row in my expression data that gene belongs to. > > I can see that this involves using symbols2indices. Is the best was > forward to simply reannotate the feature names of my expression data as > symbols and then run symbols2indices? > > e.g. > > symbols <- unlist(mget(featureNames(rmaData)) > > if rmaData is my normalised expression set. Then > > symbolsOfficial <- alias2SymbolTable(symbols,species="Hs") > > #get the genesets > load("human_c2.rdata") > #ok now set up the index of where these genes are in the genesets > c2 = symbols2indices(Hs.gmtl.c2, symbols) > > Is the c2 object now my first input to romer? > > Thanks for any advice. > > iain ______________________________________________________________________ The information in this email is confidential and intend...{{dropped:4}}