Systematic data-driven modeling of cellular systems for experimental design and hypothesis evaluation

Despite the rapid growth in biological information, drug development is still limited by our ability to understand complex cell systems that are involved in diseases. The need to model complex cell processes at many different scales requires (a) quantitative measurements from experimental techniques in cell biology and (b) mathematical modeling approaches to organize and structure information, test and validate hypotheses for the new phenomena. The goal of my thesis work was to maximally obtain and interpret information on the dynamic behavior of cellular systems using an optimal combination of computer simulations and experimental observations. I focused on two realistic biological problems to develop and demonstrate methods to generate high quality hypotheses through mathematical modeling to define the optimal next step for experimental and clinical trial design: (1) I developed a hybrid stochastic model to simulate microtubule dynamics and anti-cancer drug effects on microtubules. Model parameters were taken directly from experimental observations, and model validation was conducted against published experimental data. Our work revealed limitations in previous experimental methods to measure microtubule kinetics. (2) I developed a quantitative biological system modeling scheme to interpret flow cytometric data, and I validated this method experimentally. To do so, I designed a novel combination of protocols, assay technologies, and data analysis techniques to obtain plausible flow cytometric data. Multiple mathematical methods and reversed engineering methods, such as parameter estimation, sensitivity analysis, nonlinear regression, Akaike Information Criterion (AIC) were integrated to use a data-driven modeling to distinguish between different biological hypotheses. Our modeling and validation methodology can be generalized to apply to other biological problems with limited data.