Evaluation of Molecular Similarity/Diversity Using Topological Fragment Spectrum

在药物设计、选择化学类似物和估计分子性质方面，正在积极发展分子相似性/多样性方法，特别是在化学结构方面。在他们的案例中，亚结构特征通常被用来定义分子间的相似性(或多样性)。然而，这种结构相似性(或多样性)的量化取决于被定义为描述符的选定的子结构集。在这里，我们研究了一种使用拓扑碎片谱(TFS)方法来定量评价药物分子结构相似性或多样性的替代方法。TFS是化合物结构的拓扑轮廓的表示，它基于对化学结构中所有可能的子结构的枚举和数值表征。在这项工作中，TFS被应用于3637个化合物的结构相似性分析，这些化合物包含大约3500个随机药物和从世界药物指数中选择的所有多巴胺类活性物质。所有的光谱都对碎片进行了表征，碎片中有五个或更少的键。在数据库上进行的相似结构搜索的计算试验表明，该方法成功地适用于药物分子的结构相似性或结构多样性的评价。我们还研究了不同的相似性函数对TFS的适用性。

The use of molecular similarity/diversity methods, especially on chemical structures, is under active development in drug design, selection of analogs for chemicals and estimation of molecular properties. In their case, substructural features are often used to define intermolecular similarity (or diversity). However, quantification of such structural similarity (or diversity) depends on the chosen set of substructures defined as the descriptors. Here we investigated about an alternative approach to the quantitative evaluation of structural similarity or diversity of drug molecules using the topological fragment spectrum (TFS) method. The TFS is a representation of the topological profile of the structure of a chemical compound, and it is based on enumeration and numerical characterization of all possible substructures from the chemical structures. In the this work, the TFS was applied to the structural similarity analysis of 3,637 chemical compounds which contain around 3500 random drugs and all of the actives of dopaminergics chosen from the World Drug Index. All the spectra were characterized for the fragments, which have five or less bonds within them. The computational trials of similar structure searching on the database suggested that our approach is successfully applicable to the evaluation of structural similarity or structural diversity of drug molecules. We also investigated various similarity functions for their suitability to the TFS.