Switched Adaptive Observer for Structure Identification in Gene Regulatory Networks

Gene regulatory networks (GRNs) perform a pivotal task in conducting cellular functions. Reconstruction of these complex networks is necessary to understand underlying mechanisms directing cellular behavior. This paper deals with structure identification for gene regulatory networks. To do this end, adaptive observer is employed to estimate unknown parameters. In this method, convergence of parameter estimation to the true values depends on persistency of excitation condition. Since many real gene networks don't satisfy this condition, we propose a new method to derive these parameters. This approach is based on introducing a switching mechanism in gene networks by using biochemical perturbations. Moreover, an adaptive observer for switching systems is designed and sufficient conditions for convergence of its parameters are derived based on stability of switching systems. Proposed adaptive observer provide exponential convergence of parameters estimation and switching mechanism improve persistency of excitation. By several simulations, it is shown that the proposed method indicates better performance in contrast to the existing methods for structure identification. Regarding the obtained results, our method leads to faster convergence, handles larger unknown parameter cases and estimates true values while other methods fail.