Evolutionary exploration of Boolean networks

Random Boolean networks (RBNs) are abstract models of gene regulatory networks that govern gene expression in cells. We have developed an evolutionary model to explore the dynamic states of random Boolean networks using heuristic optimization methods. The generic behavior of random Boolean networks is investigated as the evolutionary process works its way through different generations, identifying attractors that have been suggested to resemble cell types. We investigate several fitness functions to tune RBNs with respect to the number of attractors and other network parameters such as excess graph, attractor cycle length, network sensitivity and average basin entropy. We show that by imposing particular constraints on the evolutionary model we can generate ensembles of more stable networks, which are less sensitive to perturbations. Therefore, we demonstrate that an evolutionary approach can be useful for the generation of RBN ensembles, that is sets of regulatory networks that share particular properties.