应用于蛋白质折叠模拟的三种新蒙特卡罗方法的比较

Comparison of Three New Monte Carlo Methods for Protein Folding Simulations

新蒙特卡罗方法是一类随机算法的统称.这类算法已被应用于蛋白质折叠的模拟计算,并取得了较好的结果.该文将并行回火与遗传算法的混合算法、群体模拟退火方法以及群体模拟退火方法与遗传算法的混合算法这3种改进的蒙特卡罗方法应用到蛋白质折叠模拟计算,并就二维网格模型比较了这3种方法搜索最小能量构象的能力以及计算了得到最小能量构象所花费的时间.计算机模拟计算的结果表明,3种方法对于短序列蛋白质折叠结构的预测都较为有效,而群体模拟退火方法与遗传算法的混合算法则比其它两种算法所花费的计算时间要少,也就更为有效.

New Monte Carlo method is a general designation of one type of stochastic arithmetic, which has been used in protein folding simulation with good results. Three new Monte Carlo methods, namely, a mixture of parallel tempering and genetic algorithm, population-oriented simulated annealing, as well as a mixture of population-oriented simulated annealing and genetic algorithm, are applied to a two dimensional lattice model for protein folding. The ability to find the lowest energy conformation and computational cost are compared. The result shows that all three methods are efficient for protein folding of short sequences, but a mixture of population-oriented simulated annealing and genetic algorithm uses less computational time than the other two methods and therefore is more effective.