1st CIBio Workshop

In many fields of life sciences, mathematical modeling and computational analysis are more and more exploited as complementary tools to experimental laboratory methods [Kitano 2002] . Thanks to this synergy, researchers can nowadays achieve a faster and in-depth understanding of biological systems. However, dynamic mathematical models require a proper parameterization (e.g., the kinetic parameters of reactions) to perform faithful simulations and lead to a better understanding of such systems. Kinetic parameters are often
difficult, or even impossible, to measure by means of experimental methodologies. This leads to the problem of Parameter Estimation (PE) [Chou 2009] , that is, the inference of parameters according to some indirect measurement (e.g., experimental time-series of chemical species concentrations).

​

The PE problem can be restated as an optimization problem: the goal is to find a parameterization (i.e., a vector of unknown real numbers) that allows to minimize the distance between the available experimental data (e.g., time-series) and the outcome of a simulation. In the case of biological systems, the multi-dimensional fitness landscape associated with this problem is generally highly multi-modal and rugged, so that traditional optimization methods (e.g., gradient descent) cannot be efficiently employed. On the contrary, computational intelligence methods for global optimization were shown to be suitable for this problem [Dräger 2009].

​

This talk will describe the functioning of some of the most efficient methodologies for the PE problem (e.g., differential evolution [Storn 1997] , evolution strategy and its variants [Hansen 2006] , particle swarm optimization [Kennedy 1995] ). In particular, being population-based meta-heuristics, such methods require a massive amount of fitness evaluations, i.e., independent simulations. I will describe how High-Performance Computing can be leveraged to strongly reduce the computational effort and the execution time [Nobile 2016, Nobile 2017a] , making such approaches feasible for scientific investigation. A simple example of a PE solution that exploits a novel, settings-free computational intelligence method (FST-PSO) [Nobile 2017b] will conclude the talk.

​

References

​

● [Chou 2009] Chou I. C. and Voit E. O., Recent developments in parameter estimation and structure identification of biochemical and genomic systems, Math. Biosci., vol. 219, pp. 57–83, 2009
● [Dräger 2009] Dräger A., Kronfeld M., Ziller M. J., Supper J., Planatscher H., and Magnus J. B., Modeling metabolic networks in C. glutamicum : a comparison of rate laws in combination with various parameter optimization strategies, BMC Syst. Biol., vol. 3, no. 5, 2009
● [Hansen 2006] Hansen, N.. The CMA evolution strategy: a comparing review. Towards a new evolutionary computation , 75-102, 2006
● [Kennedy 1995] Kennedy J. and Eberhart R. C., Particle swarm optimization, in Proc. IEEE Int. Conf. Neural Networks , vol. IV, Piscataway, NJ, 1995, pp. 1942–1948
● [Kitano 2002] Kitano, H.: Systems biology: a brief overview. Science, vol. 295, no. 5560, pp. 1662-1664, 2002
● [Nobile 2016] Nobile M.S., Cazzaniga P., Tangherloni A., Besozzi D.: Graphics Processing Units in Bioinformatics, Computational Biology and Systems Biology. Brief. Bioinform., vol. 18, no. 5, pp. 870-885, 2016
● [Nobile 2017a] Nobile M.S., Mauri G.: Accelerated analysis of biological parameters
space using GPUs, In: Malyshkin V. (eds) Parallel Computing Technologies. PaCT 2017. Lecture Notes in Computer Science, vol 10421, Nizhni Novgorod, Russia, 2017
● [Nobile 2017b] Nobile M.S., Cazzaniga P., Besozzi D., Colombo R., Mauri G., Pasi G.: Fuzzy Self-Tuning PSO: a settings-free algorithm for global optimization. Swarm Evol. Comp., 2017
● [Storn 1997] Storn, R., & Price, K.: Differential evolution–a simple and efficient heuristic for global optimization over continuous spaces. J. Glob. Optim., vol. 11, no. 4, pp. 341–359, 1997