A Reliable Alternative of OptKnock for Desirable Mutant Microbial Strains Conference Paper uri icon


  • © 2016 IEEE. Genome-scale metabolic models have lead to the development of computational methods to understand the underlying mechanism about the metabolism of an organism. With the foundation of Flux balance analysis (FBA), optimization methods are developed to suggest optimal knockout strategies to achieve the maximum production of the target metabolite of interest including industrial and pharmaceutical chemicals. Many of these existing methods employ bi-level optimization formulations to maximize the production of the desired biochemical at the outer-level while the organism still maintains living modeled as the inner-level optimization model, for example, by maximizing biomass production yield under the constraints imposed by gene knockouts as in the well-known OptKnock. OptKnock always selects the most optimistic flux for the production of the desired biochemical under the maximum biomass condition although the actual flux can be less than this maximum. This is because optimal knockout strategies derived in such a bi-level optimization framework may heavily depend on the closeness and robustness of the inner-level optimization model in capturing actual cell survival states. By examining the non-uniqueness of the surviving mutants from the inner-level optimization model and the fact that there may be a non-cooperative environment between the inner-level and outer-level decision makers, the suggested mutants produced by OptKnock may be different from the actual knockout strategies in practice. In order to prevent these limitations of the optimistic OptKnock, we propose a modification of it through a novel pessimistic bi-level optimization perspective. We compute pessimistic knockout solutions and compare with those from OptKnock on a core E. coli metabolic network, and observe that the solutions through pessimistic view are more reliable and perform better. For future research, we believe that the proposed pessimistic bi-level optimization model will lead to more practical and robust knockout strategies.

author list (cited authors)

  • Apaydin, M., Zeng, B. o., & Qian, X.

citation count

  • 4

publication date

  • February 2016