Seminar on "Optimization modeling in cellular biochemical networks"

The cell, as the basic unit of life, is indeed a factory consisting of thousands of biomolecules organized in a modular manner that are responsible for various highly complicated metabolic processes to transform nutrients in the environment into biomass for self-replication. With the technological advancements of genome sequencing, the metabolic capability of an organism can be captured in silico by reconstructing the so-called genome-scale metabolic network using the genome sequence from which the participating metabolic reactions and biochemicals (also called metabolites) can be inferred. Based on the mathematical representation of the metabolic network as a convex polytope and an optimality principle governing the cellular objective (e.g., maximum biomass production), linear programming, mixed integer linear programming, bilevel programming, etc. can be widely used to model the cellular metabolism and predict engineering strategies for desirable metabolic outcomes (e.g., overproduction of a chemical). Using this approach, we analyzed the overproduction of the buttery flavor compound diacetyl in a lactic acid bacterium, which was experimentally verified. We also studied the regulation of the utilization of the sugar fructose by various enzymes in the metabolic network of Corynebacterium glutamicum, which is the workhorse in the industrial production of the amino acid lysine. In another example, by integrating economical considerations, the theoretical profitability of a bacterial strain engineered to produce a target protein was optimized by choosing the optimal composition of nutrients supplied. While the metabolic modeling approach for single organisms has become mature, efforts have been increasingly put on modeling microbial communities given their omnipresence and high relevance to human health. We have recently proposed an algorithm to estimate the abundance profile of a microbial community given the nutrient condition under suitable assumptions. When applied to a gut microbial community model, the algorithm was able to predict a distribution resembling experimental data. These case studies show the applicability of optimization modeling to molecular biological networks.