With the rapid proliferation of genomic data for a variety of organisms, genome-scale simulation of metabolism is becoming an increasingly popular method of study. Often, however, there is a lack of sufficient data to solve the mathematical models generated for metabolism. By postulating an objective function representing the goal of the organism, it is possible to utilize an optimization strategy to determine the ideal distribution of metabolic resources. Even under these circumstances, often it is not possible to find a feasible solution to the optimization problem, in large part due to conflicting or incomplete constraints. Using genetic algorithms, it has been possible to develop a strategy by which key problematic constraints may be identified. Specifically, metabolic models are bred to determine the minimum number of constraints needed to be dropped in order to generate a viable metabolic model. To test the approach, a genome-scale model was developed for the obligate intracellular pathogen, Rickettsia prowazekii, the etiological agent of epidemic typhus. An objective function of maximization of ATP was chosen and justified based on the close relationship of the bacteria to mitochondria, as well as recent published studies showing the utility of the ATP maximization objective function. Simulation of the full model resulted in no feasible solution. After carrying 12 simulations involving 200 generations each, it was shown that by dropping two metabolic constraints, a viable metabolic model was generated. The constraints in question were on L-alanine and 2-keto-isovalerate. The results of the genetic algorithm suggest that L-alanine and 2-keto-isovalerate may be participating in currently unaccounted for reactions.

Extended Abstract Status: Not Uploaded