QPSO算法和罚函数在代谢通量评估中的应用

Quantum-behaved Particle Swarm Optimization Algorithm and Penalty Function in the Application of the Metabolic Flux Estimation

针对代谢通量的评估问题属于带约束的优化问题,提出了使用罚函数(penalty function,PF)的量子粒子群优化(Quan-tum-behaved particle swarm optimization,QPSO)算法来解决上述问题,通过处罚约束条件建立单一的目标函数,把带约束的化学计量转换成无约束的,然后使用QPSO算法最小化内部代谢通量。算法可用于评估谷氨酸棒杆菌(Corynebacterium glu-tamicum)的内部代谢通量,实验结果表明,提出的算法能够以较快的收敛速度找到较好的接近最优点的量化值。

Metabolic flux estimation corresponds to a constrained optimization problem. In this paper, we pro- posed a Quantum- behaved particle swarm optimization （QPSO） with penalty function to solve13C -based metabolic flux estimation problem. The stoichiometric constraints were transformed to an unconstrained one, by penalizing the constraints and building a single objective function, which in turn was minimized using QPSO algorithm for flux quan- tification. The proposed algorithm was applied to estimate the central metabolic fluxes of Corynebacterium glutami- cum. Experimental results illustrate that our algorithm is capable of achieving fast convergence to good near - optima.