基于人工智能和互联网时代的化合物毒性预测徐涛1秦越1*单伟1贾辰阳2朱加良1邓志光1李卓玥1刘丹会1(1中国核动力研究设计院核反应堆系统设计技术重点实验室成都610213;2华中师范大学化学学院农药与化学生物学教育部重点实验室武汉430079)*联系人,秦越E-mail:qyqingyue@formail.com2021-06-08收稿,2022-08-09修回,2022-09-01接受摘要随着商品中所含各种化合物的不断使用,人们日益关注其对人类及生态环境的安全危害。在过去的几年里,通过计算方法预测化合物毒性已经显示出极大的潜力。在此,总结了常用的机器学习和深度学习算法在建立毒性预测模型上的优缺点,并系统回顾了近三年发表的可免费访问的毒性预测网络服务器。此外,还讨论了基于人工智能和互联网时代下毒性预测所面临的机遇和挑战。希望指导人们合理选择算法和网络服务器进行建模及化合物毒性评估。关键词人工智能深度学习机器学习毒性预测网络服务器ToxicityPredictionBasedonArtificialIntelligenceandtheInternetEraXuTao1,QinYue1*,ShanWei1,JiaChenyang2,ZhuJialiang1,DengZhiguang1,LiZhuoyue1,LiuDanhui1(1ScienceandTechnologyonReactorSystemDesignTechnologyLaboratory,NuclearPowerInstituteofChina,Chengdu,610213;2KeyLaboratoryofPesticide&ChemicalBiology,MinistryofEducation,CollegeofChemistry,CentralChinaNormalUniversity,Wuhan,430079)AbstractWiththecontinuoususingcompoundscontainedincommodities,thereisgrowingconcernabouttheirharmtohumanandecologicalenvironmentsafety.Inthepastfewyears,computationaltechniqueshaveshowedtheirpotentialtopredicttoxicityofcompounds.Here,wesummarizetheadvantagesanddrawbacksofmachinelearninganddeeplearningalgorithmsforestablishingtoxicitypredictionmodels,andsystematicallyreviewthefreelyaccessibletoxicitypredictionwebserversforinsilicotoxicitypredictioninthepastthreeyears.Additionally,theopportunitiesandchallengesoftoxicitypredictionbasedonartificialintelligenceandinternetarediscussed.Itishopedthatthispapercanprovidehelpinguidingpeopletorationallychoosealgorithmsandwebserversformodelingandtoxicityevaluation.KeywordsArtificialintelligence,Deeplearning,Machinelearning,Toxicityprediction,Webserver“OneHealth”理念的提出,让人...