Investigation of coevolutionary approach in gene regulatory network inference

Inference of gene regulatory networks is currently an active field of research in system biology. Evolutionary computation algorithms are lately applied for finding the optimal parameters of models. This paper presents a comparison of four evolutionary algorithms (DE, GA, PSO and the hybrid Hooke-Jeeves GA) used with a linear time-variant gene network model. The paper also investigates the efficiency of cooperative coevolution approach to cope with the increased complexity of networks with large number of genes. Experiments were performed on two artificially generated and one real microarray data set. The results are twofold: the efficiency comparison may serve as a guideline for future research, and the application of coevolution proved to be successful for most algorithms.