A searching method for interacting chemical compounds for target proteins using 3D structures of compound-protein complexes

我们开发了kcombu程序来计算化合物二维结构的一对一原子对应关系。该方法基于经典的最大公共子结构（MCS）概念。该方法采用快速启发式“构建”算法，除传统的连接和断开MCS外，还计算拓扑约束断开MCS （TD-MCS）。在此基础上，我们开发了一种基于模板的化合物结构预测方法。我们还研究了与同一家族的蛋白质结合的化合物的共同和特定的3D特征。

We developed the program kcombu to calculate one-to-one atomic correspondences from 2D structures of chemical compounds. The method is based on the classical concept, maximum common substructure (MCS). Our method employs the fast heuristic “build-up” algorithm, and calculates topologoically-constrained disconnected MCS (TD-MCS) in addition to conventional connected and disconnected MCS. Based on this program, we developed a template-based structural prediction method of chemical compounds. We also researched common and specific 3D features of chemical compounds, which bind to proteins of the same family.