A Bayesian Approach to Inferring Gene Activity States from Gene Expression Data

Presented by
Craig Disselkoen
Abstract

A generally accepted model of gene activity suggests that most genes are 
either in an active or inactive state at any time/in any given condition.  
Downstream analyses of gene expression data are highly dependent upon the ability 
to correctly classify a given gene as active or inactive. Current state of the art 
methods for determining gene activity state based on observed measures of gene 
expression make limiting assumptions and do not capture statistical uncertainty. 
We will present and discuss innovative new approaches which address limitations in 
current methods.