Search
Question: GSEA with one class metaanalysis
1
gravatar for Gordon Smyth
11.8 years ago by
Gordon Smyth35k
Walter and Eliza Hall Institute of Medical Research, Melbourne, Australia
Gordon Smyth35k wrote:
Dear Mark, If I understand your problem correctly, neither GSEA nor GSA will accomodate it. The only option I know of is geneSetTest() in the limma package. This generally works well, although it will give you someone over optimistic p-values if there are strong positive correlations between the genes in your gene sets. Best wishes Gordon >Date: Fri, 02 Mar 2007 12:04:09 -0500 >From: Mark W Kimpel <mwkimpel at="" gmail.com=""> >Subject: [BioC] GSEA with one class metaanalysis >To: Bioconductor Newsgroup <bioconductor at="" stat.math.ethz.ch=""> > >I am in the process of performing a meta-analysis on multiple MA studies >run on several platforms and several species. Although I realize there >are probably other ways to approach the data, my collaborators have >chosen to use Cohen's D-statistic to summarize their data. > >I would like to use GSEA to look for over-represented groups or pathways >in the data, but what I will end up with is essentially a one-class >experiment. In other words, I would like to identify groups whose >summary D-statistic (I recognize from the recent literature that there >is more than one summary method available) is different from zero. > > From what I have read, the GSEA method is usually used to evaluate 2 >class data, the GSA method of Tibshurani can be used to evaluate >multi-class data, but can either method be easily adapted to evaluate >one class data? It would seem theoretically reasonable, but I don't know >if I would have to modify one of the packages or if one already provides >for this. > >And, while I'm at it, I will also want to look for individual genes >whose average D-statistic is different from zero. I am advocating using >a t-test with n=number of experiments with corresponding degrees of >freedom, whereas others are advocating using a z-test because the >D-statistics are summarizing many underlying individual statistics. > >Help with these questions would be deeply appreciated. > >Mark > >-- >Mark W. Kimpel MD >Neuroinformatics >Department of Psychiatry >Indiana University School of Medicine
ADD COMMENTlink written 11.8 years ago by Gordon Smyth35k
Please log in to add an answer.

Help
Access

Use of this site constitutes acceptance of our User Agreement and Privacy Policy.
Powered by Biostar version 2.2.0
Traffic: 255 users visited in the last hour