A Bayesian modeling framework for genetic regulation

Abstract

We propose an integrated framework for model creation, execution, and validation in the context of modeling genetic regulatory networks. At the center of our framework is an executable model based on Bayesian networks (BNs). We use microarray data to infer how the expression of a gene is affected by all of the other genes. We create an execution model that predicts how the system will respond to a stimulus given an initial state. Our framework is validated using a Correct Answer Known Evaluator (CAKE). CAKE also allows us to investigate how much data and what kinds of data are needed to deduce the underlying interactions.