Metabolic Design And Engineering Through Ant Colony Optimization

Due to the vast search space of all possible combinations of reaction knockouts in Escherichia coli, determining the best combination of knockouts for over-production of a metabolite of interest is a computationally expensive task. Ant colony optimization (ACO) applied to genome-scale metabolic models via flux balance analysis (FBA) provides a means by which such a solution space may feasibly be explored. In previous work, the Minimization of Metabolic Adjustment (MoMA) objective function for FBA was used in conjunction with ACO to identify the best gene knockouts for succinic acid production. In this work, algorithmic and biological constraints are introduced to further reduce the solution space. We introduce Stochastic Exploration Edge Reduction Ant Colony Optimization, or STEER-ACO. Algorithmically, ACO is modified to refine its search space over time allowing for greater initial coverage of the solution space while ultimately honing on a high quality solution. Biologically, a heuristic is introduced allowing the maximum number of knockouts to be no greater than five. Beyond this number, cellular viability becomes suspect. Results using this approach versus previous methods are reported.