An iterative parameter estimation method for biological systems

Abstract

One difficulty in building a mechanistic model of biological systems lies in determining the correct parameter values. This paper proposes a novel parameter estimation method to infer unknown parameters, such as kinetic rates, from noisy experimental observations. Derived from the Approximate Bayesian Computation (ABC) algorithm, our method can predict the distribution of each parameter rather than a single value. The Sequential Monte Carlo (SMC) method is used in this paper to approximate the real distribution of each parameter via several intermediate distributions. In order to improve the performance of the ABC SMC method, this paper develops a windowing method to reduce the parameter search space that needs to be explored. Moreover, an adaptive sampling weight which is inversely proportional to the distance value is proposed in this paper to further increase the efficiency of ABC SMC parameter estimation.