Monte-Carlo method-based QSAR model to discover phytochemical urease inhibitors using SMILES and GRAPH descriptors

Urease inhibitors are known to play a vital role in the field of medicine as well as agriculture. Special attention is attributed to the development of novel urease inhibitors with a view to treat the <i>Helicobacter pylori</i> infection. Amongst a number of urease inhibitors, a large number of molecules fail <i>in vivo</i> and in clinical trials due to their hydrolytic instability and toxicity profile. The search for potential inhibitors may require screening of large and diverse databases of small molecules and to design novel molecules. We developed a Monte-Carlo method-based QSAR model to predict urease inhibiting potency of molecules using SMILES and GRAPH descriptors on an existing diverse database of urease inhibitors. The QSAR model satisfies all the statistical parameters required for acceptance as a good model. The model is applied to identify urease inhibitors among the wide range of compounds in the phytochemical database, NPACT, as a test case. We combine the ligand-based and structure-based drug discovery methods to improve the accuracy of the prediction. The method predicts pIC₅₀ and estimates docking score of compounds in the database. The method may be applied to any other database or compounds designed <i>in silico</i> to discover novel drugs targeting urease. Communicated by Ramaswamy H. Sarma

脲酶抑制剂（urease inhibitor）在医学与农业领域均发挥着至关重要的作用。开发新型脲酶抑制剂以治疗幽门螺杆菌（Helicobacter pylori）感染，是当前备受关注的研究方向。在众多脲酶抑制剂中，大量分子因水解不稳定性与毒性谱问题，无法通过体内（in vivo）实验与临床试验。筛选潜在脲酶抑制剂通常需要对大规模多样化的小分子数据库进行筛查，并设计全新分子。本研究基于蒙特卡洛方法（Monte-Carlo method）构建了定量构效关系（QSAR）模型，利用现有多样化脲酶抑制剂数据库中的SMILES与图描述符（GRAPH descriptors），预测分子的脲酶抑制活性。该QSAR模型满足优良预测模型所需的全部统计参数标准。本研究以植物化学数据库NPACT作为测试案例，应用该模型从该数据库的海量化合物中筛选潜在脲酶抑制剂。为提升预测准确性，本研究结合了基于配体与基于结构的药物发现方法。该方法可预测化合物的pIC₅₀值，并估算数据库中化合物的对接得分（docking score）。本方法可推广至其他数据库或经计算机模拟（in silico）设计的化合物，用于发现靶向脲酶的新型药物。本文由Ramaswamy H. Sarma供稿。