基于随机森林模型的人类免疫缺陷病毒1亚型整合酶抑制剂的虚拟筛选  

Virtual Screening of HIV-1 Integrase Inhibitors Based on Random Forest Model

作　　者：王震霆 卢熙奎 伍湘平 黄凌 于慧敏 李靖阳 肖坚[2,3] WANG Zhen-ting;LU Xi-kui;WU Xiang-ping;HUANG Ling;YU Hui-min;LI Jing-yang;XIAO Jian(School of Pharmacy,Dali University,Dali Yunnan 671000,China;Department of Pharmacy,Xiangya Hospital,Central South University,Changsha 410000,China;Institute for Rational and Safe Medication Practices,National Clinical Research Center for Geriatric Disorders,Xiangya Hospital,Central South University,Changsha 410008,China)

机构地区：[1]大理大学药学院,云南大理671000 [2]中南大学湘雅医院药学部,长沙410000 [3]中南大学湘雅医院国家老年疾病临床医学研究中心老年合理用药与安全用药研究室,长沙410008

出　　处：《抗感染药学》2022年第9期1248-1255,共8页Anti-infection Pharmacy

基　　金：2021年湖南省自然科学基金(编号:2021JJ31043);2020年度国家老年疾病临床医学研究中心适宜技术推广项目(编号:XYYYJSTG-15);并行与分布处理国防科技重点实验室稳定支持项目(编号:WDZC20205500121);2021年中南大学研究生自主探索创新项目(编号:2021zzts1040);长沙市2020年度自然科学基金(编号:kq2007039)

摘　　要：目的:基于计算机辅助药物设计技术中的随机森林模型,筛选出具有潜在成药性和有人类免疫缺陷病毒1亚型(human immunodeficiency virus type 1,HIV-1)整合酶抑制活性的化合物,为HIV-1整合酶抑制剂的开发提供参考。方法:使用关键词从ChEMBL、BindingDB数据库检索出与HIV-1整合酶相关的小分子的数据资料,借助MOE软件、Knime软件等工具构建筛选HIV-1整合酶抑制剂的随机森林模型,并对从ChemDiv数据库中筛选出的小分子进行相似度验证,最后使用ADMET lab工具进行成药性分析。结果:从ChEMBL和Binding DB数据库中收集到5459个具有IC_(50)值的小分子,通过随机森林模型和相似性搜索筛选出1267个小分子,将其与作为配体的多替拉韦进行分子对接,得到24个最有小分子,再以ADMET lab软件对其进行成药性分析,最终得到5个成药性较好的小分子,其中最优的2个分子各项吸收指标均较佳,且清除率高、毒性较低。结论:通过随机森林模型的筛选和后续软件工具的虚拟验证,共得到5个具有较好成药性的HIV-1整合酶抑制剂。Objective:To screen compounds with potential druggability and human immunodeficiency virus type 1(HIV-1)integrase inhibitory activity based on the random forest model in computer-aided drug design technology,and to provide reference for the development of HIV-1 integrase inhibitors.Methods:The data of small molecules related to HIV-1 integrase was searched from the databases of ChEMBL and BindingDB using the keywords,and a random forest model for screening HIV-1 integrase inhibitors was constructed using the software MOE,Knime,etc.The small molecules screened from the ChemDiv database were verified for similarity,and finally the druggability analysis was performed using the ADMET lab tool.Results:A total of 5459 small molecules with IC_(50)values were collected from the databases of ChEMBL and BindingDB,and 1267 small molecules were screened out by random forest model and similarity search.Then molecular docking was performed with Dolutegravir as a ligand,to obtain 24 optimal small molecules.Afterwards,the druggability analysis was performed with the software ADMET lab.Finally,5 small molecules with good druggability were obtained,among which 2 optimal molecules had better absorption indexes,with a high clearance rate and a low toxicity.Conclusion:Through the screening with the random forest model and the subsequent virtual verification with software tools,5 HIV-1 integrase inhibitors with good druggability are obtained.

关 键 词：人类免疫缺陷病毒1亚型整合酶抑制剂 随机森林模型 虚拟筛选 分子对接 成药性 

分 类 号：R965.1[医药卫生—药理学]