(173h) Advanced Dynamic Flux Balance Analysis Modeling with Parametric Programming: Xdfba Modeling

Flux balance analysis (FBA) has been employed to compute flux distributions of metabolism for various types of organisms given a hypothesis in the form of objective function. FBA calculations are based on mass balances, thermodynamic constraints, and boundary conditions including exchange fluxes between the cell and environment. Recently FBA is further extended to simulate dynamic responses of a cell where FBA is computed at each discretized time with a new set of boundary conditions. The ‘slow-scale’ cell phenotype is computed by integrating macroscopic kinetic model, e.g., cell growth dynamics such as Monod equation, given the updated flux distribution at each time point. This means, however, a large scale linear programming (LP) needs to be solved at each time step, which is computationally prohibitive for genome scale FBA model.

This study proposes a parametric programming (PP) technique, which has recently found many applications in the field of advanced control for on-line optimization of large-sized system, to be incorporated into dynamic simulation of cell’s phenotype using FBA. The proposed approach, referred to as xDFBA (explicit dynamic flux balance analysis) maps intracellular fluxes as several explicit functions of substrate uptake rates and incorporates them into macroscopic kinetic model.

Case studies for Saccharomyces cerevisiae (iND750) under aerobic, and batch operations and Escherichia coli (iJR940) under anaerobic, and (fed-) batch operations were conducted to illustrate the validity and performance improvement of the proposed xDFBA approach. Compared to the online solution approach, xDFBA could remarkably reduce the computation time about 99%, which is suitable for coupling FBA analysis with macroscopic online optimization of bioreactor system such as model predictive control.