Identifying the shared metabolic objectives of glycerol bioconversion in Klebsiella pneumoniae under different culture conditions.

This paper addresses the problem of identifying the shared metabolic objectives of glycerol bioconversion in Klebsiella pneumoniae for production of 1,3-propanediol (1,3-PD) under different culture conditions. To achieve this goal, we propose a multi-level programming model. This model includes three optimization problems, where the constraint region of the first level problem is implicitly determined by the other two optimization problems. The optimized objectives of the first and second level problems are to minimize the set of fluxes that are of major importance to glycerol metabolism and the difference between the observed fluxes and those computed by the model, respectively. The third level problem in the proposed multi-level programming simultaneously solves a set of flux balance analysis (FBA) models. A method is proposed to solve efficiently the presented multi-level programming problem. In this method, we first transform the proposed multi-level problem into a bi-level problem by applying the dual theory of linear programming to the FBA models of the third level. Next, the optimal solution of the above bi-level problem is obtained by iteratively solving a sequence of mixed integer programming problems. Optimization results reveal that the proposed method can identify the shared metabolic objectives of glycerol bioconversion in Klebsiella pneumoniae under three groups of experimental data.

Multi-objective optimization of a continuous bio-dissimilation process of glycerol to 1, 3-propanediol.

This paper deals with multi-objective optimization of continuous bio-dissimilation process of glycerol to 1, 3-propanediol. In order to maximize the production rate of 1, 3-propanediol, maximize the conversion rate of glycerol to 1, 3-propanediol, maximize the conversion rate of glycerol, and minimize the concentration of by-product ethanol, we first propose six new multi-objective optimization models that can simultaneously optimize any two of the four objectives above. Then these multi-objective optimization problems are solved by using the weighted-sum and normal-boundary intersection methods respectively. Both the Pareto filter algorithm and removal criteria are used to remove those non-Pareto optimal points obtained by the normal-boundary intersection method. The results show that the normal-boundary intersection method can successfully obtain the approximate Pareto optimal sets of all the proposed multi-objective optimization problems, while the weighted-sum approach cannot achieve the overall Pareto optimal solutions of some multi-objective problems.