Gamma rankings for clustering genes in multi-group expression analysis.


In genomics, and possibly other domains of high-dimensional statistics, it can be useful to know the probabilities that N independent Gamma-distributed random variables attain each of their N! possible orderings.  Each ordering event is equivalent to an event regarding independent negative-binomial random variables, and this finding guides a dynamic-programming solution. Gamma-rank probabilities are central to a new clustering method for multi-group microarray data analysis, which I will discuss, demonstrate, and compare to alternative strategies in several examples. 

This is joint work with my PhD student Lisa M. Chung.