Reduced surface: An efficient way to compute molecular surfaces
Reduced surface: An efficient way to compute molecular surfaces
复制标题
DOI：
10.1002/(sici)1097-0282(199603)38:3
复制
发表时间：
1996-03-01
期刊：
影响因子：
2.9
通讯作者：
Spehner, JC
中科院分区：
文献类型：
--
作者：
Sanner, MF;Olson, AJ;Spehner, JC
由于其在分子模拟中的广泛应用，近年来分子表面的计算方法受到了广泛的关注。然而，大多数算法只计算溶剂可及表面的解析表示。有一个FEU程序可以计算溶剂排斥表面的解析表示，但它们在处理自相交表面的奇异情况时经常遇到问题，并且在大分子（超过10，000个原子）上往往会失败。我们在这里描述一个程序称为MSMS，这是快速和可靠的计算分子表面。它依赖于简化表面的使用，简化表面在这里被简单地定义，并且从该简化表面计算溶剂可及表面和溶剂排除表面。描述了构成MSMS的四种算法，并分析了它们的复杂度。特别注意的是给予处理的溶剂排除表面称为奇点的自交部分。并与Connolly的PQMS程序[M. L. Connolly（1993）Journal of Molecular Graphics，Vol. 11，pp. 139-141]对取自Brookhaven数据库的一组709个bl分子。MSMS能够为集合中的每个分子计算拓扑正确的表面。此外，计算表面所花费的实际时间与程序的理论复杂度一致，对于n个原子，其复杂度为O[n log（n）]。在Hewlett-Packard 9000/735工作站上，MSMS仅需0.73 s即可生成用于填充的排除溶剂的三角形表面（1crn，46个残基，327个原子，4772个三角形），嗜热菌蛋白酶为4.6 s（3 tln，316个残基，2437个原子，26462个三角形），和谷氨酰胺合成酶的104.53 s（2gls，5676个残基，43632个原子，476665个三角形）。(C)John Wiley & Sons，Inc.
Because of their wide use in molecular modeling, methods to compute molecular surfaces have received a lot of interest in recent years. However, most of the proposed algorithms compute the analytical representation of only the solvent-accessible surface. There are a feu programs that compute the analytical representation of the solvent-excluded surface, but they often have problems handling singular cases of self-intersecting surfaces and tend to fail on large molecules (more than 10,000 atoms). We describe here a program called MSMS, which is shown to be fast and reliable in computing molecular surfaces. It relies on the use of the reduced surface that is briefly defined here and from which the solvent-accessible and solvent-excluded surfaces are computed. The four algorithms composing MSMS are described and their complexity is analyzed. Special attention is given to the handling of self-intersecting parts of the solvent-excluded surface called singularities. The program has been compared with Connolly's program PQMS [M. L. Connolly (1993) Journal of Molecular Graphics, Vol. 11, pp. 139-141] on a set of 709 bl molecules taken from the Brookhaven Data Base. MSMS was able to compute topologically correct surfaces for each molecule in the set. Moreover, the actual time spent to compute surfaces is in agreement with the theoretical complexity of the program, which is shown to be O[n log(n)] for n atoms. On a Hewlett-Packard 9000/735 workstation, MSMS takes 0.73 s to produce a triangulated solvent-excluded surface for crambin (1crn, 46 residues, 327 atoms, 4772 triangles), 4.6 s for thermolysin (3tln, 316 residues, 2437 atoms, 26462 triangles), and 104.53 s for glutamine synthetase (2gls, 5676 residues, 43632 atoms, 476665 triangles). (C) 1996 John Wiley & Sons, Inc.