Cosgrove D A, Bayada D M, Johnson A P
AstraZeneca, Macclesfield, Cheshire, UK.
J Comput Aided Mol Des. 2000 Aug;14(6):573-91. doi: 10.1023/a:1008167930625.
A novel shape-based method has been developed for overlaying a series of molecule surfaces into a common reference frame. The surfaces are represented by a set of circular patches of approximately constant curvature. Two molecules are overlaid using a clique-detection algorithm to find a set of patches in the two surfaces that correspond, and overlaying the molecules so that the similar patches on the two surfaces are coincident. The method is thus able to detect areas of local, rather than global, similarity. A consensus overlay for a group of molecules is performed by examining the scores of all pairwise overlays and performing a set of overlays with the highest scores. The utility of the method has been examined by comparing the overlaid and experimental configurations of 4 sets of molecules for which there are X-ray crystal structures of the molecules bound to a protein active site. Results for the overlays are generally encouraging. Of particular note is the correct prediction of the 'reverse orientation' for ligands binding to human rhinovirus coat protein HRV14.
已开发出一种基于形状的新方法,用于将一系列分子表面叠加到一个共同的参考框架中。这些表面由一组曲率近似恒定的圆形面片表示。使用团检测算法叠加两个分子,以找到两个表面中对应的一组面片,然后叠加分子,使两个表面上相似的面片重合。因此,该方法能够检测局部而非全局的相似区域。通过检查所有成对叠加的分数并执行一组得分最高的叠加,来对一组分子进行一致性叠加。通过比较4组分子的叠加构型与实验构型(这些分子具有与蛋白质活性位点结合的X射线晶体结构),检验了该方法的实用性。叠加结果总体上令人鼓舞。特别值得注意的是,对于与人类鼻病毒衣壳蛋白HRV14结合的配体,正确预测了其“反向取向”。