(210c) Evolutionary Algorithms in Metabolic Flux Analysis: The Case for Optimizing Bio-Butanol Production in Clostridium Acetobutylicum

The anaerobe Clostridium acetobutylicum is of considerable interest for the conversion of complex and diverse substrates to bio-butanol in high titers. Genome-scale metabolic flux models have recently been published for this organism, and here we present new results illustrating the use of evolutionary (e.g., genetic) algorithms in effectively predicting dominant phenotypes. Decoupling the metabolic programs of acidogenesis (e.g., acetate, butyrate, and lactate production) during exponential growth from solventogenesis (e.g., acetone, butanol, and ethanol production) in the stationary phase presents a unique and important challenge to the systems biologist. We previously introduced the role of genetic algorithms in reducing the size of the phenotypic solution space based on proton influx/efflux to aid convergence of the C. acetobutylicum genome-scale model. In addition, genetic algorithms were used to suggest the dynamic characteristics of the biomass constituting equation with this organism. Here, this concept is explored in much greater detail and is supported with experimental results. In turn, we show how a dynamic biomass constituting equation can be used to uncover regulatory relationships. We also developed the concept of numerically-determined sub-systems using metabolite flux ratios to clarify the possible flux solutions about a known singularity in the metabolic network. The usefulness of real-coded genetic algorithms is demonstrated in this approach for (i) identifying metabolite flux ratios that define singularities and (ii) solving the resulting systems of equations. Comparisons with traditional linear programming solutions in the presence and absence of constraints reveal significant advantages of the evolutionary programming approach at the expense of computational resources.