BioE PhD Proposal Presentation- Justin Lee

Advisor:

Mark Styczynski, Ph.D. (Georgia Institute of Technology - ChBE)

Committee:

Fani Boukouvala, Ph.D. (Georgia Institute of Technology - ChBE)

Melissa Kemp, Ph.D. (Georgia Institute of Technology - BME)

Andrew Medford, Ph.D. (Georgia Institute of Technology - ChBE)

Eberhard Voit, Ph.D. (Georgia Institute of Technology - BME)

Computational Modeling of Metabolic Pathways Toward Predicting Dynamic Phenotypes

Predicting phenotypes of perturbations in a metabolic network is one of the longstanding goals of systems biology and is critical for metabolic engineering applications. Two of the most common frameworks used in systems biology to model and study metabolic networks are constraint-based models and ordinary differential equation (ODE) based models, which both have advantages and disadvantages. While constraint-based models are simple and easily solvable, they lack the details necessary to measure metabolite dynamics. ODE-based models are much more intricate and are able to track metabolite concentrations, but require kinetic parameters that are often unknown and difficult to estimate. Our group has recently created Linear Kinetics-Dynamic Flux Balance Analysis (LK-DFBA), which combines the efficiency of constraint-based models and the ability to calculate metabolite dynamics from ODE-based models. While LK-DFBA has been shown to recapitulate the data generated from the original system it was fitted to, it is unknown if LK-DFBA is capable of predicting previously unobserved phenotypes after system perturbations. In this proposal, I have devised steps to improve the predictive power of LK-DFBA by using novel linear constraints that are more representative of interactions found in biological systems. I will also develop a method for identifying allosteric regulatory interactions that are critical to the LK-DFBA modeling framework and are often unknown for systems that are not well-studied. Finally, I will create a systematic approach for inferring absolute concentration values from raw metabolomics data, which will allow metabolomics data to be more seamlessly integrated into LK-DFBA.