Abstract
The inference of underlying genetic networks from the observed time-series data of gene expression patterns has become one of the major topics in the bioinformatics fields. The S-system model is considered an ideal choose to inferring genetic networks because it is rich enough in structure to capture various dynamics and some methods are available for analyzing it. However, the number of S-system parameters is proportional to the square of the number of genes. This is why inference algorithms based on the S-system model have only been applied to small or medium scale networks. This paper uses a hybrid evolutionary algorithm to optimize the parameters. Moreover, using the idea of problem decomposition strategy to resolve the high-dimensionality of the genetic network inference problem. The optimization problem is first decomposed into several sub-problems. Each sub-problem is solved using SSO. After that, the solutions of sub-problems are combined and use ABC to solve the original optimization problem. To verify our proposed method, we conduct three experiments. Moreover, we compare the PSO and ABC between four various target networks. The result shows that SSO is efficient enough to solve sub-problems and the proposed hybrid evolutionary algorithm performs well in inferring large-scale S-system models.