[BioC] GSEA for meta-analysis

Mark W Kimpel mwkimpel at gmail.com
Fri Mar 2 18:52:00 CET 2007


My apology if this comes across as a duplicate post, my mail server 
crashed just as I was sending a few minutes ago.

I am in the process of performing a meta-analysis on micro-array data 
from 7 different experiments; several platforms and two species were 
used. My collaborators have sent me the Cohen's D-statistic for their 
experiments, which I will integrate with those of mine.

I would like to use GSEA to look for groups with higher than expected 
average Cohen's D-stats. This should be theoretically possible as GSEA 
usually uses t- for F-statistics. Can either of the packages available 
(GSEA or GSA) accomadate such one-class data without modification?

We also wish to identify individual genes whose mean d-statistic is not 
zero. I have suggested using a t-test with n=number of experiments with 
the corresponding degrees of freedom whereas others have suggested we do 
a z-test, since the d-statistics are already summaries of many 
individual animals. Which would be correct?

Thanks,

Mark

-- 
Mark W. Kimpel MD
Neuroinformatics
Department of Psychiatry
Indiana University School of Medicine



More information about the Bioconductor mailing list