Parameter Estimation for Biochemical Pathways using Swarm AlgorithmTan Chee Meng1, Sandeep Somani2
email@example.com, Bioinformatics Institute; firstname.lastname@example.org, Bioinformatics Institute
Systems Biology provides a holistic view of the interaction and dynamics between various biochemical pathways in the cell. Advancement in genomics, proteomics and metabolomics has helped generate large amount of data, enabling reverse engineering of gene regulation and metabolic pathways. An analytical model consists of quantitative information on biological components and their reaction kinetics. However due to paucity of quantitative data, various numerical optimization techniques have been employed. The accuracy of kinetic parameter estimation largely depends on the type of approach used and the available search space Some of the parameter estimation techniques include (a) deterministic approaches like Levenberg-Marquardt (LM) algorithm, Sequential Quadratic Programming, and (b) stochastic approach like Simulated Annealing, Genetic Algorithms and Evolutionary Algorithms. All these approaches share a common goal i.e., to find a global optimal solution. Considering that biological systems exhibit high robustness and operate at a broad range of kinetic parameters, it is important that optimization techniques capture this important property of cells. In this study, we present an optimization technique called SWARM. The main advantage of SWARM lies in its capability of detecting multiple optimal solutions. SWARM has been successfully tested on several benchmark mathematical examples described in the poster. The next step would be to apply SWARM on well-characterized biochemical pathways before moving on to a more challenging domain i.e., to unravel robustness of the system and discover unknown kinetics of the interactions.