基于多层量子粒子群优化算法和Toy模型的蛋白质结构预测研究  

Research of protein structure prediction based on multiple-layer quantum-behaved particle swarm optimization and toy model

作　　者：李成渊[1] 丁颜蕊[1] 须文波[1] 

机构地区：[1]江南大学物联网工程学院,江苏无锡214122

出　　处：《计算机与应用化学》2010年第11期1574-1578,共5页Computers and Applied Chemistry

基　　金：江南大学青年创新团队(JNIRT0702))

摘　　要：蛋白质结构预测,作为计算生物学基本问题之一,是个典型的NP难解问题。研究表明合理运用算法,借助物理模型,可用于预测蛋白质结构。Toy模型就是较为简单的一类,其势能最低状态的确定则为结构预测的关键所在。量子粒子群算法是典型的智能优化算法,已广泛应用于多种系统寻优问题中。本篇文章提出使用1种改进的量子粒子群优化算法,并结合Toy模型,进行蛋白质结构预测。算法的改进在于对每次迭代的粒子,排序之后将种群分成精英子群、开采子群和勘探子群来区别处理,并通过实验进行运算和预测。结果表明运用改进的量子粒子群优化算法来进行蛋白质折叠结构预测是可行的且高效的。Protein structure prediction,known as an NP-complete problem,is one of the basic problems in computational biology.Studies show that physical models are available to be applied,reasonable use of algorithms,to predict protein structure.Toy model is one of the relatively simple physical models,while the determination of ground state is the key for structure prediction.Quantum particle swarm optimization is a typical of intelligent optimization algorithm,has been widely used in a variety of system optimization problems.This article made use of an improved quantum particle swarm optimization to find the ground state and predict the protein structure.Improvement lies in each iteration,after sorting the particles are divided into elite subgroup,exploitation subgroup and exploration subgroup between sub-groups to deal with,and operation through experiments and prediction between sub-groups to deal with,sequentially leading to predict.The results show that the use of improved quantum particle swarm optimization with the Toy model,structure prediction is feasible and efficient.

关 键 词：量子粒子群算法 Toy模型 蛋白质结构预测 

分 类 号：TP301.6[自动化与计算机技术—计算机系统结构]