We present a novel method for parameter sensitivity identification based on the Covariance Matrix Adaptation Evolution Strategy (CMA-ES) optimization procedure. Our method directly uses the information that is acquired during an optimization process to provide local and global parameter sensitivity estimates with no additional evaluations of the cost function. To demonstrate the performance of the method, we consider a complex example from cell biology, namely a TCR-activated signal transduction network. For this network, different global sensitivity indices are known.
The CMA-ES is a powerful evolutionary algorithm for non-linear, non-convex optimization problems. It has successfully been applied in a variety of domains ranging from lens design in optics over computation of Nash equilibria in economics to design of cancer chemotherapies. Moreover, the CMA-ES has demonstrated unique performance in robust parameter estimation for biochemical network models.