Probabilistic Approximation and Analysis Techniques for Bio-Pathway Models

Quantitative modeling of bio-pathway dynamics is crucial to the system-level understanding of cellular functions and behavior. Currently, a common method of representing bio-pathways is through a system of ordinary differential equations (ODEs). However, calibrating and analyzing large ODE-based pathway models often requires a large number of numerical simulations. To address this issue, this thesis presents an approximation approach. This consists of discretization of time and value space, sampling of a prior distribution of initial states, numerical simulations and suitable counting leading to a dynamic Bayesian network. Consequently tasks such as parameter estimation and global sensitivity analysis can be efficiently carried out through standard Bayesian inference techniques. We have demonstrated the applicability of our techniques by studying two existing pathways taken from Brown et al. and Goldbeter et al., and a "live" pathway called the complement system in collaboration with Ding et al. Apart from improved performance, our method matches the lack of precision and noise in the experimental data and produces probabilistic estimates. In addition, the crucial insights we have gained from the study of complement system could contribute to the development of immunomodulation therapies.

Author dc:creator

• LIU BING

• Computational Systems Biology, Bio-pathway Modeling, Ordinary Differential Equations, Dynamic Bayesian Networks, Innate Immunity, C4BP