摘要
研究了生物信息学中的一个重要问题 ,即蛋白质结构预测 受物理世界的物体间相互作用的规律的启发 ,给出了该问题一个二维欧氏空间连续模型 它比离散模型有一定的优越性 ,此模型的优点可能在于让计算很自然地利用到了一个客观存在的“天然导引” ,这个“天然导引”即是疏水氨基酸之间的引力 ,从而在构形优度相当的前提下 ,连续模型有助于计算速度的提高 然后根据这个连续模型找到了相应的拟物算法 ,最后给出了一些实验结果 。
An important problem in biology informatics, that is, protein structure prediction, is studied in this paper. Enlightened by the law of reciprocity among things in the physics world, a continuous model in 2D Euclidean space is given which gain an advantage over the discretion model in some aspects. The continuous model's strong point is probably that it utilizes an external “inartificial leading” which is the gravitation among hydrophobe amino acid. Accordingly, in the presupposition of the equivalent result, the continuous model redounds to improving the computing rapidity. Then, a related quasiphysical algorithm is formulated according to the 2D continuous model. Finally some experiment results are given, which also prove the advantages of the continuous model and its quasiphysical algorithm.
出处
《计算机研究与发展》
EI
CSCD
北大核心
2004年第11期1959-1965,共7页
Journal of Computer Research and Development
基金
国家"九七三"重点基础研究发展规划基金项目 (G19980 3 0 60 0 )
关键词
蛋白质结构预测
NP难度问题
折叠
拟物算法
引力势能
protein structure prediction
NP-hard problem
folding
quasiphysical algorithm
attraction potential energy
