Proceeding talk – Theme: Systems.
Abstract
Approximate Bayesian Computation coupled to Sequential Monte Carlo (ABC-SMC) algorithms constitute a powerful approach for parameter estimation and model selection of mathematical models of complex biological systems. A crucial step in the ABC-SMC algorithms, significantly affecting their performance, is the propagation of a set of parameter vectors through a sequence of intermediate distributions. We employ Dirichlet process mixtures (DPMs) to design optimal transition kernels and we present an ABC-SMC algorithm with DPM kernels. We illustrate the use of the proposed methodology using real data for the canonical Wnt signalling pathway. The results indicate that DPM kernels are more efficient in the exploration of the parameter space and can significantly improve ABC-SMC performance providing better estimations of complex multimodal distributions. The method is used to estimate the parameters and the initial state of two models of the Wnt pathway and a novel multicompartment model is shown to fit better the data.
Authors
Konstantinos Koutroumpas, Laboratory MICS, CentraleSupélec, Chatenay-Malabry, 92295, France
Paolo Ballarini, Laboratory MICS, CentraleSupélec, Chatenay-Malabry, 92295, France
Irene Votsi, Laboratory MICS, CentraleSupélec, Chatenay-Malabry, 92295, France
Paul-Henry Cournède, Laboratory MICS, CentraleSupélec, Chatenay-Malabry, 92295, France