Revealing structure of complex biological systems using bayesian networks

Abstract

Bayesian networks represent statistical dependencies among variables; they are able to model multiple types of relationships, including stochastic, non-linear, and arbitrary combinatoric. Such flexibility has made them excellent models for reverse-engineering structure of complex networks. This chapter reviews the use of Bayesian networks for probing structure of biological systems. We begin with an introduction to Bayesian networks, addressing especially issues of their interpretation as relates to understanding system structure. We then cover how Bayesian network structures are learned from data, considering a popular scoring metric, the BDe, in detail. We finish by reviewing the uses of Bayesian networks in biological systems to date and the concurrent advances in Bayesian network methodology tailored for use in biology. © 2010 Springer-Verlag London Limited.