Estimating the Unmeasured Dynamics of Biological Systems using a Constrained Real-Coded Genetic Algorithm

Optimization is critical in the development and implementation of techniques for estimating the unmeasured dynamics of biological systems. Local-search algorithms present an attractive option for implementing this optimization due to their low complexity. These algorithms can, however, produce sub-optimal results when estimating these unmeasured dynamics using noisy measurements. In this work, we implement a global-search constrained optimization based on real-coded genetic algorithms (RC-GA) to estimate the unmeasured dynamics of the Caspases signal transduction network. We show that RC-GA performs better than gradient-based local search algorithms by providing estimates that more closely match the "true" dynamics of the biological system.