求解蛋白质结构预测问题的局部搜索算法

蛋白质结构预测问题是计算生物学领域的核心问题之一。通过理论计算的方法根据蛋白质氨基酸序列直接预测其空间结构是解决这一问题的有效途径。构造了新的邻域结构，采用了部分随机跳坑策略，对此问题提出了新的局部搜索算法。计算结果表明，该算法计算效率要优于传统的遗传算法和Monte Carlo方法。对于链长为50的算例还找到了文献中所没有的全新的最低能量构形。

Local Search Algorithm for Solving Protein Structure Prediction Problem

Protein structure prediction has proven to be one of the central problems in the field of computational biology. It is a feasible approach to predict theoretically the three-dimensional structure of proteins based only on amino acid sequence information. Using a new neighborhood structure and partly randomized off-trap strategy, a novel local search algorithm for protein structure prediction is proposed. Computational results demonstrate that our algorithm not only is more efficient than conventional genetic and Monte Carlo algorithms, but also find new configurations of lowest energy states missed in orevious papers for the sequence of length N=50