多模态虚拟筛选方法筛选靶向抑制冠状病毒Nsp5蛋白的天然分子药物

打开文本图片集

中图分类号：Q811.4 文献标志码：A

Multimodal Virtual Screening Method for Identifying Natural Molecule Drugs Targeting Inhibition of Coronavirus Nsp5 Protein

WANG Chenyu ， XU Xinyia， ZHANG Haoa，HU Haifeng ， WU Jiansheng

（a. School of Chemistry and Life Sciences，b. School of Communications and Information Engineering， c.SchoolofComputerScience，NanjingUniversityofPostsandTelecommunications，Nanjing21oo23，Jiangsu，China）

Abstract：To screen potential inhibitors targeting Nsp5 protein of severe acute respiratory syndrome coronavirus （SARSCoV-2）froma natural compound database，anew multimodal virtual screening method was developed by integrating onedimensional and thre-dimensional molecular information.The candidate compounds obtained through thisscreening process can inhibit Nsp5 protein and serve asa foundation for developing drugs to treatcoronavirus disease（COVID19）.SuperNatural database was preliminarily screened using dropout additive regression trees（DART）algorithm with a random nactivation mechanism.The docking scores of the molecules were calculated using molecular docking software， and the top three ranked molecules were validated by molecular dynamics simulations to confirm their dynamic binding efects.Theresults showthatthecandidate molecules obtainedbythe proposed method exhibit diversityandachieve higher docking scores compared tounimodal methods.Molecular dynamics simulations show thatthe selected high-scoring molecules demonstrate good dynamic docking effects at the dynamic binding site of the target protein.

Keywords：multi-modal virtual screening method;severe acute respiratory syndrome coronavirus；Nsp5 protein；natural compound;molecular dynamics simulations

冠状病毒病（C0VID-19）是由严重急性呼吸综合征冠状病毒（ SARS-CoV-2; 引起的疾病[1]。（剩余16114字）