A computationally efficient solution strategy for optimal gene knockouts for targeted overproduction
Conference Paper
Overview
Identity
Additional Document Info
Other
View All
Overview
abstract
Copyright is held by the author/owner(s). A recently developed adaptive bi-level optimization algorithm, MOMAKnock, has the objective to maximize the targeted biochemical overproduction as the outer-level problem, and MOMA criterion modeling the survival of mutants as the inner-level objective function. This method gives improved targeted overproductions with more robust knockout strategies. An adaptive solution with piecewise linear approximation to the inner-level objective function has been employed in the original work. In this project, Karush-Kuhn-Tucker (KKT) conditions are used to convert bi-level MOMAKnock to a single level mixed integer programming problem since the inner-level problem is a convex optimization problem. We compare our KKT-based solution strategy with the original adaptive solution strategy by evaluating both strategies on a small E.coli core metabolic network. The experimental results show that our new KKT-based solution is computationally more efficient, and achieves orders of magnitude speedup to obtain the optimal solutions. Copyright is held by the author/owner(s).
name of conference
Proceedings of the 6th ACM Conference on Bioinformatics, Computational Biology and Health Informatics
