Hi all,
To identify the biological relevance of a set of candidate genes (from differentially expression analysis or clustering or any other method ) the first and foremost thing that comes into mind is to perform an "enrichment analysis", something along the lines of a hypergeometric test or fisher chi square test or a bit more sophisticated approaches like Gene Set Enrichment Analysis (GSEA). I have been searching around to see if somebody has worked on alternative approach to give a "biological meaning" to a set of candidate genes, but with very little success. Have you come across any alternative approaches ? what are your thoughts on this ?
"My problem with these methods is that the statistical combinatorial space is not representative of the possible biological combinatorial space." I completely agree with you here.
Isn't reducing the combinatorial space as you said more suited to a wet lab experiment ? may be some sort of functional genomics assay to understand the function of a gene ? I have a feeling that enrichment analysis tests like hypergeometric tests are highly subjective, and quite often enrichment results for the same set of candidate gene can be made "more biologically relevant" or "less biologically relevant" just by playing around with the number of universal genes selected (total number of white + black balls in the urn).