Willett P
Krebs Institute for Biomolecular Research, University of Sheffield, UK.
J Mol Recognit. 1995 Sep-Oct;8(5):290-303. doi: 10.1002/jmr.300080503.
This paper provides an overview of the research that has been carried out in Sheffield over the last decade into searching techniques for databases of three-dimensional (3D) chemical structures. A 3D structure or query pattern is represented by a labelled graph, in which the nodes and the edges of the graph are used to represent atoms and the associated inter-atomic distances, respectively. The presence of a pharmacophore in each of the structures in a database can then be tested by means of a subgraph isomorphism algorithm, the computational requirements of which are minimized by the use of an initial screening procedure that eliminates the majority of the structures from the subgraph-isomorphism search. Analogous graph-based representation and searching methods can also be used with flexible 3D structures: in this case, the edges of the graphs represent inter-atomic distance ranges and a final conformational search needs to be carried out for those molecules that match the query pharmacophore in the subgraph-isomorphism search. The paper also reviews related work on the automatic identification of pharmacophoric patterns and on 3D similarity searching.
本文概述了过去十年间在谢菲尔德开展的关于三维（3D）化学结构数据库搜索技术的研究。三维结构或查询模式由带标签的图表示，其中图的节点和边分别用于表示原子和相关的原子间距离。然后可以通过子图同构算法来测试数据库中每个结构中是否存在药效团，通过使用初始筛选程序将子图同构搜索中大部分结构排除，可将该算法的计算需求降至最低。基于图的类似表示和搜索方法也可用于柔性三维结构：在这种情况下，图的边表示原子间距离范围，对于在子图同构搜索中与查询药效团匹配的那些分子，需要进行最终的构象搜索。本文还回顾了关于药效团模式自动识别和三维相似性搜索的相关工作。
