Agent-based optimization of biological response networks

One of the major challenges in systems biology today is to devise robust methods of interpreting data concerning the expression levels of the genes in an organism in a way that will shed light on the collective interactions between multiple genes and their products. The ability to better understand and predict the structures and actions of complex biological systems is of significant importance to modern drug discovery as well as our understanding of the mechanisms behind an organism's ability to react to its environment. In this paper we present a study for robust biological pathway construction through an agent-based methodology-probability collectives multi-agent systems (PCMAS). This technique relies on the search of particular seed nodes by probability collectives to construct biological networks based on various sets of interaction information and gene expression data. As an application, expression data of ofloxacin response in M. tuberculosis is used to build response networks. We then demonstrate how this approach provides robust prediction of response networks to facilitate drug target identification on systems-level.