GPU-powered evolutionary design of mass-action-based models of gene regulation

This chapter introduces a computational strategy to infer a reaction-based models (RBM) that specifically represents a gene regulation model (GRM) characterized by some predefined behavior. It then presents a two-level evolutionary design (ED) methodology, named cuGENED, which integrates two evolutionary computation (EC) algorithms, namely Cartesian genetic programming (CGP) and particle swarm optimization (PSO). The chapter also describes the formalization of GRMs by means of mass-action-based models, and gives a brief introduction of the EC techniques used in the ED of gene regulatory networks (GRNs). Then, it briefly explains the graphics processing units (GPUs) computing framework exploited to speed up the optimization process. The chapter further provides a detailed description of the ED methodology to automatically derive GRMs. Finally, it presents the results of the application of cuGENED for the automatic design of GRNs consisting in two and three genes