The inference of genetic regulatory networks from time-course data is one of the main challenges in systems biology. The ultimate goal of inferred model is to obtain the expressions quantitatively comprehending every detail and principle of biological systems. This study introduces a multiobjective optimization approach to infer a realizable S-system structure for genetic regulatory networks. The work of inference is to minimize simultaneously the concentration error, slope error and interaction measure in order to find a suitable S-system model structure and its corresponding model parameters. Hybrid differential evolution is applied to solve the ?-constrained problem, which is converted from the multiobjective optimization problem, for minimizing the interaction measure with subject to the expectation constraints for the concentration and slope error criteria. This approach could avoid assigning a suitable penalty weight for sum of magnitude of kinetic orders for the penalty problem in order to prune the model structure.
Date:
2008-06
Relation:
IEEE Congress on Evolutionary Computation. 2008 Jun:1736-1743.
