Article(id=1198656155770782448, tenantId=1146029695717560320, journalId=1189982191388893191, issueId=1198656143976399200, articleNumber=null, orderNo=null, doi=10.16438/j.0513-4870.2022-1453, pmid=null, cstr=null, oa=null, hot=null, price=null, onlineType=0, articleFormat=0, articleType=null, articleTypeStr=null, receivedDate=1672675200000, receivedDateStr=2023-01-03, revisedDate=1676217600000, revisedDateStr=2023-02-13, acceptedDate=null, acceptedDateStr=null, onlineDate=1763711497489, onlineDateStr=2025-11-21, pubDate=1694448000000, pubDateStr=2023-09-12, doiRegisterDate=null, doiRegisterDateStr=null, onlineIssueDate=1763711497489, onlineIssueDateStr=2025-11-21, onlineJustAcceptDate=null, onlineJustAcceptDateStr=null, onlineFirstDate=null, onlineFirstDateStr=null, sourceXml=null, magXml=null, createTime=1763711497489, creator=13701087609, updateTime=1763711497489, updator=13701087609, issue=Issue{id=1198656143976399200, tenantId=1146029695717560320, journalId=1189982191388893191, year='2023', volume='58', issue='9', pageStart='2541', pageEnd='2834', issueExtLink='null', onlineDate='null', pubDate='1694448000000', pubDateStr='2023-09-12', beforeIssueId=null, nextIssueId=null, price=null, status=1, issueComplete=1, articleOrder=1, issueType=-1, specialIssue=null, createTime=1763711494677, creator='13701087609', updateTime=1763711620095, updator='13701087609', preIssue=null, nextIssue=null, articleTotal=null, ext={EN=IssueExt(id=1198656670072144034, tenantId=1146029695717560320, journalId=1189982191388893191, issueId=1198656143976399200, language=EN, specialIssueTitle=, coverIllustrator=null, specialIssueEditor=, specialIssueAbout=), CN=IssueExt(id=1198656670072144035, tenantId=1146029695717560320, journalId=1189982191388893191, issueId=1198656143976399200, language=CN, specialIssueTitle=, coverIllustrator=null, specialIssueEditor=, specialIssueAbout=)}, issueFiles=null, downloadFileDto=null}, startPage=2610, endPage=2622, ext={EN=ArticleExt(id=1198656156546728743, articleId=1198656155770782448, tenantId=1146029695717560320, journalId=1189982191388893191, language=EN, title=Deep learning-based de novo drug design: principles, tools and practice, columnId=null, journalTitle=Acta Pharmaceutica Sinica, columnName=null, runingTitle=null, highlight=null, articleAbstract=

Design of structurally-novel drug molecules with deep learning can overcome the technical bottleneck of classical computer-aided drug design. It has become the frontier of new technique research on drug design, and has shown great potential in drug research and development practice. This review starts from the basic principles of deep learning-driven de novo drug design, goes on with the brief introduction to deep molecular generation techniques as well as computational tools and the analysis on representative successful cases, and eventually provides our perspective for future direction and application prospect about this technique. This review will provide ideas on new technique research and references for new drug research and development practice to which this technique is applied.

, authors=null, authorsList=Tao SHEN, Dong-mei WANG, Song WU, Jian-dong JIANG, Jie XIA, authorCompany=null, correspAuthors=Jie XIA, authorNote=null, correspAuthorsNote=null, copyrightStatement=Copyright ©2023 Acta Pharmaceutica Sinica. All rights reserved., copyrightOwner=null, extLink=null, articleAbsUrl=null, sourceXml=null, magXml=null, pdfUrl=null, pdf=null, pdfFileSize=null, pdfExtLink=null, richHtmlUrl=null, mobilePdfUrl=null, reviewReport=null, pdfFirstPage=null, abstractGraph=null, abstractGraphContent=null, abstractVideo=null, citation=null, cebUrl=null, magXmlContent=null, mapNumber=null, fund=null), CN=ArticleExt(id=1198656160573259875, articleId=1198656155770782448, tenantId=1146029695717560320, journalId=1189982191388893191, language=CN, title=基于深度学习的全新药物分子设计: 原理、工具与实践, columnId=1190335349655180086, journalTitle=药学学报, columnName=综述, runingTitle=null, highlight=null, articleAbstract=

利用深度学习方法设计具有全新结构的药物分子可以突破传统计算机辅助药物设计的技术瓶颈, 已经成为药物设计新技术研究的前沿, 在药物研发实践中展现了巨大潜力。本文从深度学习驱动的全新分子设计的基本原理出发, 简要介绍了深度分子生成技术和计算工具, 分析了若干代表性成功案例, 最后对该技术的未来发展方向和应用前景进行了展望。本综述将为新技术研究提供思路, 为应用该技术开展新药研发实践提供借鉴。

, authors=null, authorsList=沈涛, 王冬梅, 吴松, 蒋建东, 夏杰, authorCompany=null, correspAuthors=夏杰, authorNote=null, correspAuthorsNote=
*夏杰, E-mail:
, copyrightStatement=版权所有©《药学学报》编辑部2023, copyrightOwner=null, extLink=null, articleAbsUrl=null, sourceXml=skDqfNpZyBiBjOD5icvkyw==, magXml=NgsUtiPIUkSsZA0ed0KM5w==, pdfUrl=null, pdf=LWAKxKjCUA9SpWCcMHf95w==, pdfFileSize=1866260, pdfExtLink=null, richHtmlUrl=null, mobilePdfUrl=null, reviewReport=null, pdfFirstPage=null, abstractGraph=YAycL5C/vgnRwrGOb/JF+g==, abstractGraphContent=null, abstractVideo=null, citation=null, cebUrl=null, magXmlContent=MUj/Bd4LUZ73Z6w9knf1RQ==, mapNumber=null, fund=null)}, authors=[Author(id=1198960249093849965, tenantId=1146029695717560320, journalId=1189982191388893191, articleId=1198656155770782448, orderNo=0, firstName=null, middleName=null, lastName=null, nameCn=null, orcid=null, stid=null, country=null, authorPic=null, dead=0, email=null, emailSecond=null, emailThird=null, correspondingAuthor=0, authorType=1, ext={EN=AuthorExt(id=1198960249228067705, tenantId=1146029695717560320, journalId=1189982191388893191, articleId=1198656155770782448, authorId=1198960249093849965, language=EN, stringName=Tao SHEN, firstName=Tao, middleName=null, lastName=SHEN, prefix=null, suffix=null, authorComment=null, nameInitials=null, affiliation=null, department=null, xref=null, address=State Key Laboratory of Bioactive Substance and Function of Natural Medicines, Institute of Materia Medica, Chinese Academy of Medical Sciences & Peking Union Medical College, Beijing 100050, China, bio=null, bioImg=null, bioContent=null, aboutCorrespAuthor=null), CN=AuthorExt(id=1198960249387451274, tenantId=1146029695717560320, journalId=1189982191388893191, articleId=1198656155770782448, authorId=1198960249093849965, language=CN, stringName=沈涛, firstName=涛, middleName=null, lastName=沈, prefix=null, suffix=null, authorComment=null, nameInitials=null, affiliation=null, department=null, xref=null, address=中国医学科学院、北京协和医学院药物研究所, 天然药物活性物质与功能国家重点实验室, 北京 100050, bio=null, bioImg=null, bioContent=null, aboutCorrespAuthor=null)}, companyList=[AuthorCompany(id=1198960248963826525, tenantId=1146029695717560320, journalId=1189982191388893191, articleId=1198656155770782448, xref=null, ext=[AuthorCompanyExt(id=1198960248972215135, tenantId=1146029695717560320, journalId=1189982191388893191, articleId=1198656155770782448, companyId=1198960248963826525, language=EN, country=null, province=null, city=null, postcode=null, companyName=null, departmentName=null, remark=State Key Laboratory of Bioactive Substance and Function of Natural Medicines, Institute of Materia Medica, Chinese Academy of Medical Sciences & Peking Union Medical College, Beijing 100050, China), AuthorCompanyExt(id=1198960248976409440, tenantId=1146029695717560320, journalId=1189982191388893191, articleId=1198656155770782448, companyId=1198960248963826525, language=CN, country=null, province=null, city=null, postcode=null, companyName=null, departmentName=null, remark=中国医学科学院、北京协和医学院药物研究所, 天然药物活性物质与功能国家重点实验室, 北京 100050)])]), Author(id=1198960249567806369, tenantId=1146029695717560320, journalId=1189982191388893191, articleId=1198656155770782448, orderNo=1, firstName=null, middleName=null, lastName=null, nameCn=null, orcid=null, stid=null, country=null, authorPic=null, dead=0, email=null, emailSecond=null, emailThird=null, correspondingAuthor=0, authorType=1, ext={EN=AuthorExt(id=1198960249731384244, tenantId=1146029695717560320, journalId=1189982191388893191, articleId=1198656155770782448, authorId=1198960249567806369, language=EN, stringName=Dong-mei WANG, firstName=Dong-mei, middleName=null, lastName=WANG, prefix=null, suffix=null, authorComment=null, nameInitials=null, affiliation=null, department=null, xref=null, address=State Key Laboratory of Bioactive Substance and Function of Natural Medicines, Institute of Materia Medica, Chinese Academy of Medical Sciences & Peking Union Medical College, Beijing 100050, China, bio=null, bioImg=null, bioContent=null, aboutCorrespAuthor=null), CN=AuthorExt(id=1198960249873990599, tenantId=1146029695717560320, journalId=1189982191388893191, articleId=1198656155770782448, authorId=1198960249567806369, language=CN, stringName=王冬梅, firstName=冬梅, middleName=null, lastName=王, prefix=null, suffix=null, authorComment=null, nameInitials=null, affiliation=null, department=null, xref=null, address=中国医学科学院、北京协和医学院药物研究所, 天然药物活性物质与功能国家重点实验室, 北京 100050, bio=null, bioImg=null, bioContent=null, aboutCorrespAuthor=null)}, companyList=[AuthorCompany(id=1198960248963826525, tenantId=1146029695717560320, journalId=1189982191388893191, articleId=1198656155770782448, xref=null, ext=[AuthorCompanyExt(id=1198960248972215135, tenantId=1146029695717560320, journalId=1189982191388893191, articleId=1198656155770782448, companyId=1198960248963826525, language=EN, country=null, province=null, city=null, postcode=null, companyName=null, departmentName=null, remark=State Key Laboratory of Bioactive Substance and Function of Natural Medicines, Institute of Materia Medica, Chinese Academy of Medical Sciences & Peking Union Medical College, Beijing 100050, China), AuthorCompanyExt(id=1198960248976409440, tenantId=1146029695717560320, journalId=1189982191388893191, articleId=1198656155770782448, companyId=1198960248963826525, language=CN, country=null, province=null, city=null, postcode=null, companyName=null, departmentName=null, remark=中国医学科学院、北京协和医学院药物研究所, 天然药物活性物质与功能国家重点实验室, 北京 100050)])]), Author(id=1198960249991431126, tenantId=1146029695717560320, journalId=1189982191388893191, articleId=1198656155770782448, orderNo=2, firstName=null, middleName=null, lastName=null, nameCn=null, orcid=null, stid=null, country=null, authorPic=null, dead=0, email=null, emailSecond=null, emailThird=null, correspondingAuthor=0, authorType=1, ext={EN=AuthorExt(id=1198960250121454570, tenantId=1146029695717560320, journalId=1189982191388893191, articleId=1198656155770782448, authorId=1198960249991431126, language=EN, stringName=Song WU, firstName=Song, middleName=null, lastName=WU, prefix=null, suffix=null, authorComment=null, nameInitials=null, affiliation=null, department=null, xref=null, address=State Key Laboratory of Bioactive Substance and Function of Natural Medicines, Institute of Materia Medica, Chinese Academy of Medical Sciences & Peking Union Medical College, Beijing 100050, China, bio=null, bioImg=null, bioContent=null, aboutCorrespAuthor=null), CN=AuthorExt(id=1198960250264060921, tenantId=1146029695717560320, journalId=1189982191388893191, articleId=1198656155770782448, authorId=1198960249991431126, language=CN, stringName=吴松, firstName=松, middleName=null, lastName=吴, prefix=null, suffix=null, authorComment=null, nameInitials=null, affiliation=null, department=null, xref=null, address=中国医学科学院、北京协和医学院药物研究所, 天然药物活性物质与功能国家重点实验室, 北京 100050, bio=null, bioImg=null, bioContent=null, aboutCorrespAuthor=null)}, companyList=[AuthorCompany(id=1198960248963826525, tenantId=1146029695717560320, journalId=1189982191388893191, articleId=1198656155770782448, xref=null, ext=[AuthorCompanyExt(id=1198960248972215135, tenantId=1146029695717560320, journalId=1189982191388893191, articleId=1198656155770782448, companyId=1198960248963826525, language=EN, country=null, province=null, city=null, postcode=null, companyName=null, departmentName=null, remark=State Key Laboratory of Bioactive Substance and Function of Natural Medicines, Institute of Materia Medica, Chinese Academy of Medical Sciences & Peking Union Medical College, Beijing 100050, China), AuthorCompanyExt(id=1198960248976409440, tenantId=1146029695717560320, journalId=1189982191388893191, articleId=1198656155770782448, companyId=1198960248963826525, language=CN, country=null, province=null, city=null, postcode=null, companyName=null, departmentName=null, remark=中国医学科学院、北京协和医学院药物研究所, 天然药物活性物质与功能国家重点实验室, 北京 100050)])]), Author(id=1198960250448609298, tenantId=1146029695717560320, journalId=1189982191388893191, articleId=1198656155770782448, orderNo=3, firstName=null, middleName=null, lastName=null, nameCn=null, orcid=null, stid=null, country=null, authorPic=null, dead=0, email=null, emailSecond=null, emailThird=null, correspondingAuthor=0, authorType=1, ext={EN=AuthorExt(id=1198960250616381478, tenantId=1146029695717560320, journalId=1189982191388893191, articleId=1198656155770782448, authorId=1198960250448609298, language=EN, stringName=Jian-dong JIANG, firstName=Jian-dong, middleName=null, lastName=JIANG, prefix=null, suffix=null, authorComment=null, nameInitials=null, affiliation=null, department=null, xref=null, address=State Key Laboratory of Bioactive Substance and Function of Natural Medicines, Institute of Materia Medica, Chinese Academy of Medical Sciences & Peking Union Medical College, Beijing 100050, China, bio=null, bioImg=null, bioContent=null, aboutCorrespAuthor=null), CN=AuthorExt(id=1198960250775765050, tenantId=1146029695717560320, journalId=1189982191388893191, articleId=1198656155770782448, authorId=1198960250448609298, language=CN, stringName=蒋建东, firstName=建东, middleName=null, lastName=蒋, prefix=null, suffix=null, authorComment=null, nameInitials=null, affiliation=null, department=null, xref=null, address=中国医学科学院、北京协和医学院药物研究所, 天然药物活性物质与功能国家重点实验室, 北京 100050, bio=null, bioImg=null, bioContent=null, aboutCorrespAuthor=null)}, companyList=[AuthorCompany(id=1198960248963826525, tenantId=1146029695717560320, journalId=1189982191388893191, articleId=1198656155770782448, xref=null, ext=[AuthorCompanyExt(id=1198960248972215135, tenantId=1146029695717560320, journalId=1189982191388893191, articleId=1198656155770782448, companyId=1198960248963826525, language=EN, country=null, province=null, city=null, postcode=null, companyName=null, departmentName=null, remark=State Key Laboratory of Bioactive Substance and Function of Natural Medicines, Institute of Materia Medica, Chinese Academy of Medical Sciences & Peking Union Medical College, Beijing 100050, China), AuthorCompanyExt(id=1198960248976409440, tenantId=1146029695717560320, journalId=1189982191388893191, articleId=1198656155770782448, companyId=1198960248963826525, language=CN, country=null, province=null, city=null, postcode=null, companyName=null, departmentName=null, remark=中国医学科学院、北京协和医学院药物研究所, 天然药物活性物质与功能国家重点实验室, 北京 100050)])]), Author(id=1198960250889011268, tenantId=1146029695717560320, journalId=1189982191388893191, articleId=1198656155770782448, orderNo=4, firstName=null, middleName=null, lastName=null, nameCn=null, orcid=null, stid=null, country=null, authorPic=null, dead=0, email=jie.william.xia@hotmail.com, emailSecond=null, emailThird=null, correspondingAuthor=1, authorType=1, ext={EN=AuthorExt(id=1198960251002257487, tenantId=1146029695717560320, journalId=1189982191388893191, articleId=1198656155770782448, authorId=1198960250889011268, language=EN, stringName=Jie XIA, firstName=Jie, middleName=null, lastName=XIA, prefix=null, suffix=null, authorComment=null, nameInitials=null, affiliation=null, department=null, xref=*, address=State Key Laboratory of Bioactive Substance and Function of Natural Medicines, Institute of Materia Medica, Chinese Academy of Medical Sciences & Peking Union Medical College, Beijing 100050, China, bio=null, bioImg=null, bioContent=null, aboutCorrespAuthor=null), CN=AuthorExt(id=1198960251098726491, tenantId=1146029695717560320, journalId=1189982191388893191, articleId=1198656155770782448, authorId=1198960250889011268, language=CN, stringName=夏杰, firstName=杰, middleName=null, lastName=夏, prefix=null, suffix=null, authorComment=null, nameInitials=null, affiliation=null, department=null, xref=*, address=中国医学科学院、北京协和医学院药物研究所, 天然药物活性物质与功能国家重点实验室, 北京 100050, bio=null, bioImg=null, bioContent=null, aboutCorrespAuthor=null)}, companyList=[AuthorCompany(id=1198960248963826525, tenantId=1146029695717560320, journalId=1189982191388893191, articleId=1198656155770782448, xref=null, ext=[AuthorCompanyExt(id=1198960248972215135, tenantId=1146029695717560320, journalId=1189982191388893191, articleId=1198656155770782448, companyId=1198960248963826525, language=EN, country=null, province=null, city=null, postcode=null, companyName=null, departmentName=null, remark=State Key Laboratory of Bioactive Substance and Function of Natural Medicines, Institute of Materia Medica, Chinese Academy of Medical Sciences & Peking Union Medical College, Beijing 100050, China), AuthorCompanyExt(id=1198960248976409440, tenantId=1146029695717560320, journalId=1189982191388893191, articleId=1198656155770782448, companyId=1198960248963826525, language=CN, country=null, province=null, city=null, postcode=null, companyName=null, departmentName=null, remark=中国医学科学院、北京协和医学院药物研究所, 天然药物活性物质与功能国家重点实验室, 北京 100050)])])], keywords=[Keyword(id=1198960251371356283, tenantId=1146029695717560320, journalId=1189982191388893191, articleId=1198656155770782448, language=EN, orderNo=1, keyword=new drug research and development), Keyword(id=1198960251505574027, tenantId=1146029695717560320, journalId=1189982191388893191, articleId=1198656155770782448, language=EN, orderNo=2, keyword=molecular informatics), Keyword(id=1198960251602043036, tenantId=1146029695717560320, journalId=1189982191388893191, articleId=1198656155770782448, language=EN, orderNo=3, keyword=de novo drug design), Keyword(id=1198960251824341166, tenantId=1146029695717560320, journalId=1189982191388893191, articleId=1198656155770782448, language=EN, orderNo=4, keyword=artificial intelligence), Keyword(id=1198960252021473471, tenantId=1146029695717560320, journalId=1189982191388893191, articleId=1198656155770782448, language=EN, orderNo=5, keyword=deep learning), Keyword(id=1198960252256354514, tenantId=1146029695717560320, journalId=1189982191388893191, articleId=1198656155770782448, language=CN, orderNo=1, keyword=新药研发), Keyword(id=1198960252361212125, tenantId=1146029695717560320, journalId=1189982191388893191, articleId=1198656155770782448, language=CN, orderNo=2, keyword=分子信息学), Keyword(id=1198960252516401384, tenantId=1146029695717560320, journalId=1189982191388893191, articleId=1198656155770782448, language=CN, orderNo=3, keyword=全新药物设计), Keyword(id=1198960252684173563, tenantId=1146029695717560320, journalId=1189982191388893191, articleId=1198656155770782448, language=CN, orderNo=4, keyword=人工智能), Keyword(id=1198960252856140045, tenantId=1146029695717560320, journalId=1189982191388893191, articleId=1198656155770782448, language=CN, orderNo=5, keyword=深度学习)], refs=[Reference(id=1198960254722605529, tenantId=1146029695717560320, journalId=1189982191388893191, articleId=1198656155770782448, doi=10.1021/jm4004285, pmid=null, pmcid=null, year=2014, volume=57, issue=null, pageStart=4977, pageEnd=5010, url=null, language=null, rfNumber=[1], rfOrder=0, authorNames=null, journalName=J Med Chem, refType=null, unstructuredReference=Cherkasov A, Muratov EN, Fourches D, et al. QSAR modeling: where have you been? where are you going to?[J]. J Med Chem, 2014, 57: 4977-5010., articleTitle=QSAR modeling: where have you been? where are you going to?, refAbstract=null), Reference(id=1198960254852628960, tenantId=1146029695717560320, journalId=1189982191388893191, articleId=1198656155770782448, doi=null, pmid=null, pmcid=null, year=2014, volume=null, issue=null, pageStart=12, pageEnd=13, url=null, language=null, rfNumber=[2], rfOrder=1, authorNames=null, journalName=Structure-based Design of Drugs and Other Bioactive Molecules: Tools and Strategies, refType=null, unstructuredReference=Ghosh AK, Gemma S. Structure-based Design of Drugs and Other Bioactive Molecules: Tools and Strategies[M]. New Jersey: John Wiley & Sons, 2014: 12-13., articleTitle=null, refAbstract=null), Reference(id=1198960255053955577, tenantId=1146029695717560320, journalId=1189982191388893191, articleId=1198656155770782448, doi=null, pmid=null, pmcid=null, year=2019, volume=49, issue=null, pageStart=1375, pageEnd=1394, url=null, language=null, rfNumber=[3], rfOrder=2, authorNames=null, journalName=Sci Sin Vit (中国科学: 生命科学), refType=null, unstructuredReference=Tan XQ, Xiong JC, Zhu TF, et al. Development of drug design in China: 40 years of achievements[J]. Sci Sin Vit (中国科学: 生命科学), 2019, 49: 1375-1394, articleTitle=Development of drug design in China: 40 years of achievements, refAbstract=null), Reference(id=1198960255200756232, tenantId=1146029695717560320, journalId=1189982191388893191, articleId=1198656155770782448, doi=null, pmid=null, pmcid=null, year=null, volume=null, issue=null, pageStart=null, pageEnd=null, url=null, language=null, rfNumber=[4], rfOrder=3, authorNames=null, journalName=null, refType=null, unstructuredReference=Aitken M, Kleinrock M. Global Trends in R & D: Overview through 2021 [EB/OL]. (2022-02-10) [2022-12-24]. https://www.iqvia.com/insights/the-iqvia-institute/reports/global-trends-in-r-and-d-2022., articleTitle=null, refAbstract=null), Reference(id=1198960255360139799, tenantId=1146029695717560320, journalId=1189982191388893191, articleId=1198656155770782448, doi=10.1016/j.metabol.2017.01.011, pmid=null, pmcid=null, year=2017, volume=69, issue=null, pageStart=S36, pageEnd=S40, url=null, language=null, rfNumber=[5], rfOrder=4, authorNames=null, journalName=Metabolism, refType=null, unstructuredReference=Hamet P, Tremblay J. Artificial intelligence in medicine[J]. Metabolism, 2017, 69: S36-S40., articleTitle=Artificial intelligence in medicine, refAbstract=null), Reference(id=1198960255615992357, tenantId=1146029695717560320, journalId=1189982191388893191, articleId=1198656155770782448, doi=10.1145/3065386, pmid=null, pmcid=null, year=2017, volume=60, issue=null, pageStart=84, pageEnd=90, url=null, language=null, rfNumber=[6], rfOrder=5, authorNames=null, journalName=Commun ACM, refType=null, unstructuredReference=Krizhevsky A, Sutskever I, Hinton GE. Imagenet classification with deep convolutional neural networks[J]. Commun ACM, 2017, 60: 84-90., articleTitle=Imagenet classification with deep convolutional neural networks, refAbstract=null), Reference(id=1198960255779570224, tenantId=1146029695717560320, journalId=1189982191388893191, articleId=1198656155770782448, doi=10.1038/nature14539, pmid=null, pmcid=null, year=2015, volume=521, issue=null, pageStart=436, pageEnd=444, url=null, language=null, rfNumber=[7], rfOrder=6, authorNames=null, journalName=Nature, refType=null, unstructuredReference=LeCun Y, Bengio Y, Hinton G. Deep learning[J]. Nature, 2015, 521: 436-444., articleTitle=Deep learning, refAbstract=null), Reference(id=1198960255959925312, tenantId=1146029695717560320, journalId=1189982191388893191, articleId=1198656155770782448, doi=10.1021/acschembio.6b00706, pmid=null, pmcid=null, year=2017, volume=12, issue=null, pageStart=23, pageEnd=35, url=null, language=null, rfNumber=[8], rfOrder=7, authorNames=null, journalName=ACS Chem Biol, refType=null, unstructuredReference=Humbeck L, Koch O. What can we learn from bioactivity data? chemoinformatics tools and applications in chemical biology research[J]. ACS Chem Biol, 2017, 12: 23-35., articleTitle=What can we learn from bioactivity data? chemoinformatics tools and applications in chemical biology research, refAbstract=null), Reference(id=1198960256077365833, tenantId=1146029695717560320, journalId=1189982191388893191, articleId=1198656155770782448, doi=null, pmid=null, pmcid=null, year=2018, volume=20, issue=null, pageStart=1878, pageEnd=1912, url=null, language=null, rfNumber=[9], rfOrder=8, authorNames=null, journalName=Brief Bioinform, refType=null, unstructuredReference=Rifaioglu AS, Atas H, Martin MJ, et al. Recent applications of deep learning and machine intelligence on in silico drug discovery: methods, tools and databases[J]. Brief Bioinform, 2018, 20: 1878-1912, articleTitle=Recent applications of deep learning and machine intelligence on in silico drug discovery: methods, tools and databases, refAbstract=null), Reference(id=1198960256333218399, tenantId=1146029695717560320, journalId=1189982191388893191, articleId=1198656155770782448, doi=10.1016/j.apsb.2022.02.002, pmid=null, pmcid=null, year=2022, volume=12, issue=null, pageStart=3049, pageEnd=3062, url=null, language=null, rfNumber=[10], rfOrder=9, authorNames=null, journalName=Acta Pharm Sin B, refType=null, unstructuredReference=Sun DX, Gao W, Hu HX, et al. Why 90% of clinical drug development fails and how to improve it?[J]. Acta Pharm Sin B, 2022, 12: 3049-3062., articleTitle=Why 90% of clinical drug development fails and how to improve it?, refAbstract=null), Reference(id=1198960256438076012, tenantId=1146029695717560320, journalId=1189982191388893191, articleId=1198656155770782448, doi=10.1016/0010-4809(73)90074-8, pmid=null, pmcid=null, year=1973, volume=6, issue=null, pageStart=411, pageEnd=421, url=null, language=null, rfNumber=[11], rfOrder=10, authorNames=null, journalName=Comput Biomed Res, refType=null, unstructuredReference=Hiller SA, Golender VE, Rosenblit AB, et al. Cybernetic methods of drug design. Ⅰ. Statement of the problem—the perceptron approach[J]. Comput Biomed Res, 1973, 6: 411-421., articleTitle=Cybernetic methods of drug design. Ⅰ. Statement of the problem—the perceptron approach, refAbstract=null), Reference(id=1198960256605848190, tenantId=1146029695717560320, journalId=1189982191388893191, articleId=1198656155770782448, doi=10.1021/acs.jcim.9b00237, pmid=null, pmcid=null, year=2019, volume=59, issue=null, pageStart=3370, pageEnd=3388, url=null, language=null, rfNumber=[12], rfOrder=11, authorNames=null, journalName=J Chem Inf Model, refType=null, unstructuredReference=Yang K, Swanson K, Jin W, et al. Analyzing learned molecular representations for property prediction[J]. J Chem Inf Model, 2019, 59: 3370-3388., articleTitle=Analyzing learned molecular representations for property prediction, refAbstract=null), Reference(id=1198960256735871632, tenantId=1146029695717560320, journalId=1189982191388893191, articleId=1198656155770782448, doi=10.1021/acs.jcim.2c01073, pmid=null, pmcid=null, year=2022, volume=62, issue=null, pageStart=5938, pageEnd=5951, url=null, language=null, rfNumber=[13], rfOrder=12, authorNames=null, journalName=J Chem Inf Model, refType=null, unstructuredReference=van Tilborg D, Alenicheva A, Grisoni F. Exposing the limitations of molecular machine learning with activity cliffs[J]. J Chem Inf Model, 2022, 62: 5938-5951., articleTitle=Exposing the limitations of molecular machine learning with activity cliffs, refAbstract=null), Reference(id=1198960256874283673, tenantId=1146029695717560320, journalId=1189982191388893191, articleId=1198656155770782448, doi=10.1038/s42256-022-00581-6, pmid=null, pmcid=null, year=2022, volume=4, issue=null, pageStart=1246, pageEnd=1255, url=null, language=null, rfNumber=[14], rfOrder=13, authorNames=null, journalName=Nat Mach Intell, refType=null, unstructuredReference=Janela T, Bajorath J. Simple nearest-neighbour analysis meets the accuracy of compound potency predictions using complex machine learning models[J]. Nat Mach Intell, 2022, 4: 1246-1255., articleTitle=Simple nearest-neighbour analysis meets the accuracy of compound potency predictions using complex machine learning models, refAbstract=null), Reference(id=1198960257021084324, tenantId=1146029695717560320, journalId=1189982191388893191, articleId=1198656155770782448, doi=10.3390/ijms22189983, pmid=null, pmcid=null, year=2021, volume=22, issue=null, pageStart=9983, pageEnd=null, url=null, language=null, rfNumber=[15], rfOrder=14, authorNames=null, journalName=Int J Mol Sci, refType=null, unstructuredReference=Kim J, Park S, Min D, et al. Comprehensive survey of recent drug discovery using deep learning[J]. Int J Mol Sci, 2021, 22: 9983., articleTitle=Comprehensive survey of recent drug discovery using deep learning, refAbstract=null), Reference(id=1198960257151107767, tenantId=1146029695717560320, journalId=1189982191388893191, articleId=1198656155770782448, doi=10.1038/s41587-019-0224-x, pmid=null, pmcid=null, year=2019, volume=37, issue=null, pageStart=1038, pageEnd=1040, url=null, language=null, rfNumber=[16], rfOrder=15, authorNames=null, journalName=Nat Biotechnol, refType=null, unstructuredReference=Zhavoronkov A, Ivanenkov YA, Aliper A, et al. Deep learning enables rapid identification of potent DDR1 kinase inhibitors[J]. Nat Biotechnol, 2019, 37: 1038-1040., articleTitle=Deep learning enables rapid identification of potent DDR1 kinase inhibitors, refAbstract=null), Reference(id=1198960257314685633, tenantId=1146029695717560320, journalId=1189982191388893191, articleId=1198656155770782448, doi=10.1021/acsmedchemlett.0c00088, pmid=null, pmcid=null, year=2020, volume=11, issue=null, pageStart=1496, pageEnd=1505, url=null, language=null, rfNumber=[17], rfOrder=16, authorNames=null, journalName=ACS Med Chem Lett, refType=null, unstructuredReference=Vanhaelen Q, Lin YC, Zhavoronkov A. The advent of generative chemistry[J]. ACS Med Chem Lett, 2020, 11: 1496-1505., articleTitle=The advent of generative chemistry, refAbstract=null), Reference(id=1198960257465680586, tenantId=1146029695717560320, journalId=1189982191388893191, articleId=1198656155770782448, doi=10.4155/fmc-2018-0358, pmid=null, pmcid=null, year=2019, volume=11, issue=null, pageStart=567, pageEnd=597, url=null, language=null, rfNumber=[18], rfOrder=17, authorNames=null, journalName=Future Med Chem, refType=null, unstructuredReference=Xu YJ, Lin KJ, Wang SW, et al. Deep learning for molecular generation[J]. Future Med Chem, 2019, 11: 567-597., articleTitle=Deep learning for molecular generation, refAbstract=null), Reference(id=1198960257650229975, tenantId=1146029695717560320, journalId=1189982191388893191, articleId=1198656155770782448, doi=10.1039/C9ME00039A, pmid=null, pmcid=null, year=2019, volume=4, issue=null, pageStart=828, pageEnd=849, url=null, language=null, rfNumber=[19], rfOrder=18, authorNames=null, journalName=Mol Syst Des Eng, refType=null, unstructuredReference=Elton DC, Boukouvalas Z, Fuge MD, et al. Deep learning for molecular design—a review of the state of the art[J]. Mol Syst Des Eng, 2019, 4: 828-849., articleTitle=Deep learning for molecular design—a review of the state of the art, refAbstract=null), Reference(id=1198960257771864801, tenantId=1146029695717560320, journalId=1189982191388893191, articleId=1198656155770782448, doi=10.1016/j.sbi.2021.10.001, pmid=null, pmcid=null, year=2022, volume=72, issue=null, pageStart=135, pageEnd=144, url=null, language=null, rfNumber=[20], rfOrder=19, authorNames=null, journalName=Curr Opin Struct Biol, refType=null, unstructuredReference=Wang M, Wang Z, Sun H, et al. Deep learning approaches for de novo drug design: an overview[J]. Curr Opin Struct Biol, 2022, 72: 135-144., articleTitle=Deep learning approaches for de novo drug design: an overview, refAbstract=null), Reference(id=1198960257952219884, tenantId=1146029695717560320, journalId=1189982191388893191, articleId=1198656155770782448, doi=10.1039/C8SC04175J, pmid=null, pmcid=null, year=2019, volume=10, issue=null, pageStart=1692, pageEnd=1701, url=null, language=null, rfNumber=[21], rfOrder=20, authorNames=null, journalName=Chem Sci, refType=null, unstructuredReference=Winter R, Montanari F, Noé F, et al. Learning continuous and data-driven molecular descriptors by translating equivalent chemical representations[J]. Chem Sci, 2019, 10: 1692-1701., articleTitle=Learning continuous and data-driven molecular descriptors by translating equivalent chemical representations, refAbstract=null), Reference(id=1198960258103214842, tenantId=1146029695717560320, journalId=1189982191388893191, articleId=1198656155770782448, doi=10.1021/ci00057a005, pmid=null, pmcid=null, year=1988, volume=28, issue=null, pageStart=31, pageEnd=36, url=null, language=null, rfNumber=[22], rfOrder=21, authorNames=null, journalName=J Chem Inf Comput Sci, refType=null, unstructuredReference=Weininger D. SMILES, a chemical language and information system. 1. Introduction to methodology and encoding rules[J]. J Chem Inf Comput Sci, 1988, 28: 31-36., articleTitle=SMILES, a chemical language and information system. 1. Introduction to methodology and encoding rules, refAbstract=null), Reference(id=1198960258237432585, tenantId=1146029695717560320, journalId=1189982191388893191, articleId=1198656155770782448, doi=10.1021/ci00062a008, pmid=null, pmcid=null, year=1989, volume=29, issue=null, pageStart=97, pageEnd=101, url=null, language=null, rfNumber=[23], rfOrder=22, authorNames=null, journalName=J Chem Inf Comput Sci, refType=null, unstructuredReference=Weininger D, Weininger A, Weininger JL. SMILES. 2. Algorithm for generation of unique SMILES notation[J]. J Chem Inf Comput Sci, 1989, 29: 97-101., articleTitle=SMILES. 2. Algorithm for generation of unique SMILES notation, refAbstract=null), Reference(id=1198960258384233240, tenantId=1146029695717560320, journalId=1189982191388893191, articleId=1198656155770782448, doi=10.1038/s42256-021-00368-1, pmid=null, pmcid=null, year=2021, volume=3, issue=null, pageStart=759, pageEnd=770, url=null, language=null, rfNumber=[24], rfOrder=23, authorNames=null, journalName=Nat Mach Intell, refType=null, unstructuredReference=Skinnider MA, Stacey RG, Wishart DS, et al. Chemical language models enable navigation in sparsely populated chemical space[J]. Nat Mach Intell, 2021, 3: 759-770., articleTitle=Chemical language models enable navigation in sparsely populated chemical space, refAbstract=null), Reference(id=1198960258526839590, tenantId=1146029695717560320, journalId=1189982191388893191, articleId=1198656155770782448, doi=10.1016/j.patter.2022.100588, pmid=null, pmcid=null, year=2022, volume=3, issue=null, pageStart=100588, pageEnd=null, url=null, language=null, rfNumber=[25], rfOrder=24, authorNames=null, journalName=Patterns, refType=null, unstructuredReference=Krenn M, Ai Q, Barthel S, et al. SELFIES and the future of molecular string representations[J]. Patterns, 2022, 3: 100588., articleTitle=SELFIES and the future of molecular string representations, refAbstract=null), Reference(id=1198960258665251633, tenantId=1146029695717560320, journalId=1189982191388893191, articleId=1198656155770782448, doi=10.1088/2632-2153/aba947, pmid=null, pmcid=null, year=2020, volume=1, issue=null, pageStart=045024, pageEnd=null, url=null, language=null, rfNumber=[26], rfOrder=25, authorNames=null, journalName=Mach Learn Sci Technol, refType=null, unstructuredReference=Krenn M, Häse F, Nigam A, et al. Self-referencing embedded strings (SELFIES): a 100% robust molecular string representation[J]. Mach Learn Sci Technol, 2020, 1: 045024., articleTitle=Self-referencing embedded strings (SELFIES): a 100% robust molecular string representation, refAbstract=null), Reference(id=1198960258828829505, tenantId=1146029695717560320, journalId=1189982191388893191, articleId=1198656155770782448, doi=10.1186/s13321-018-0287-6, pmid=null, pmcid=null, year=2018, volume=10, issue=null, pageStart=33, pageEnd=null, url=null, language=null, rfNumber=[27], rfOrder=26, authorNames=null, journalName=J Cheminform, refType=null, unstructuredReference=Li Y, Zhang L, Liu Z. Multi-objective de novo drug design with conditional graph generative model[J]. J Cheminform, 2018, 10: 33., articleTitle=Multi-objective de novo drug design with conditional graph generative model, refAbstract=null), Reference(id=1198960258988213066, tenantId=1146029695717560320, journalId=1189982191388893191, articleId=1198656155770782448, doi=10.3389/fphar.2020.565644, pmid=null, pmcid=null, year=2020, volume=11, issue=null, pageStart=565644, pageEnd=null, url=null, language=null, rfNumber=[28], rfOrder=27, authorNames=null, journalName=Front Pharmacol, refType=null, unstructuredReference=Polykovskiy D, Zhebrak A, Sanchez-Lengeling B, et al. Molecular sets (MOSES): a benchmarking platform for molecular generation models[J]. Front Pharmacol, 2020, 11: 565644., articleTitle=Molecular sets (MOSES): a benchmarking platform for molecular generation models, refAbstract=null), Reference(id=1198960259122430811, tenantId=1146029695717560320, journalId=1189982191388893191, articleId=1198656155770782448, doi=10.1016/j.drudis.2022.03.017, pmid=null, pmcid=null, year=2022, volume=27, issue=null, pageStart=2209, pageEnd=2215, url=null, language=null, rfNumber=[29], rfOrder=28, authorNames=null, journalName=Drug Discov Today, refType=null, unstructuredReference=Pillai N, Dasgupta A, Sudsakorn S, et al. Machine learning guided early drug discovery of small molecules[J]. Drug Discov Today, 2022, 27: 2209-2215., articleTitle=Machine learning guided early drug discovery of small molecules, refAbstract=null), Reference(id=1198960259290202982, tenantId=1146029695717560320, journalId=1189982191388893191, articleId=1198656155770782448, doi=10.1021/acscentsci.7b00572, pmid=null, pmcid=null, year=2018, volume=4, issue=null, pageStart=268, pageEnd=276, url=null, language=null, rfNumber=[30], rfOrder=29, authorNames=null, journalName=ACS Cent Sci, refType=null, unstructuredReference=Gómez-Bombarelli R, Wei JN, Duvenaud D, et al. Automatic chemical design using a data-driven continuous representation of molecules[J]. ACS Cent Sci, 2018, 4: 268-276., articleTitle=Automatic chemical design using a data-driven continuous representation of molecules, refAbstract=null), Reference(id=1198960259445392240, tenantId=1146029695717560320, journalId=1189982191388893191, articleId=1198656155770782448, doi=null, pmid=null, pmcid=null, year=null, volume=null, issue=null, pageStart=null, pageEnd=null, url=null, language=null, rfNumber=31, rfOrder=30, authorNames=null, journalName=null, refType=null, unstructuredReference=Guimaraes GL, Sanchez-Lengeling B, Outeiral C, et al. Objective-reinforced generative adversarial networks (ORGAN) for sequence generation models [J]. arXiv, 2017: 1705.10843., articleTitle=null, refAbstract=null), Reference(id=1198960259629941630, tenantId=1146029695717560320, journalId=1189982191388893191, articleId=1198656155770782448, doi=10.1021/acscentsci.7b00512, pmid=null, pmcid=null, year=2018, volume=4, issue=null, pageStart=120, pageEnd=131, url=null, language=null, rfNumber=[32], rfOrder=31, authorNames=null, journalName=ACS Cent Sci, refType=null, unstructuredReference=Segler MHS, Kogej T, Tyrchan C, et al. Generating focused molecule libraries for drug discovery with recurrent neural networks[J]. ACS Cent Sci, 2018, 4: 120-131., articleTitle=Generating focused molecule libraries for drug discovery with recurrent neural networks, refAbstract=null), Reference(id=1198960259806102408, tenantId=1146029695717560320, journalId=1189982191388893191, articleId=1198656155770782448, doi=10.1186/s13321-017-0235-x, pmid=null, pmcid=null, year=2017, volume=9, issue=null, pageStart=48, pageEnd=null, url=null, language=null, rfNumber=[33], rfOrder=32, authorNames=null, journalName=J Cheminform, refType=null, unstructuredReference=Olivecrona M, Blaschke T, Engkvist O, et al. Molecular de-novo design through deep reinforcement learning[J]. J Cheminform, 2017, 9: 48., articleTitle=Molecular de-novo design through deep reinforcement learning, refAbstract=null), Reference(id=1198960260003234714, tenantId=1146029695717560320, journalId=1189982191388893191, articleId=1198656155770782448, doi=10.1039/D0SC03126G, pmid=null, pmcid=null, year=2020, volume=11, issue=null, pageStart=8312, pageEnd=8322, url=null, language=null, rfNumber=[34], rfOrder=33, authorNames=null, journalName=Chem Sci, refType=null, unstructuredReference=Yang Y, Zheng S, Su S, et al. SyntaLinker: automatic fragment linking with deep conditional transformer neural networks[J]. Chem Sci, 2020, 11: 8312-8322., articleTitle=SyntaLinker: automatic fragment linking with deep conditional transformer neural networks, refAbstract=null), Reference(id=1198960260145841059, tenantId=1146029695717560320, journalId=1189982191388893191, articleId=1198656155770782448, doi=null, pmid=null, pmcid=null, year=null, volume=null, issue=null, pageStart=null, pageEnd=null, url=null, language=null, rfNumber=35, rfOrder=34, authorNames=null, journalName=null, refType=null, unstructuredReference=Wu L, Gong C, Liu X, et al. Diffusion-based molecule generation with informative prior bridges [J]. arXiv, 2022: 2209.00865., articleTitle=null, refAbstract=null), Reference(id=1198960260384916403, tenantId=1146029695717560320, journalId=1189982191388893191, articleId=1198656155770782448, doi=10.1038/s42256-022-00448-w, pmid=null, pmcid=null, year=2022, volume=4, issue=null, pageStart=180, pageEnd=186, url=null, language=null, rfNumber=[36], rfOrder=35, authorNames=null, journalName=Nat Mach Intell, refType=null, unstructuredReference=Godinez WJ, Ma EJ, Chao AT, et al. Design of potent antimalarials with generative chemistry[J]. Nat Mach Intell, 2022, 4: 180-186., articleTitle=Design of potent antimalarials with generative chemistry, refAbstract=null), Reference(id=1198960260548494269, tenantId=1146029695717560320, journalId=1189982191388893191, articleId=1198656155770782448, doi=10.1039/D1SC05976A, pmid=null, pmcid=null, year=2022, volume=13, issue=null, pageStart=2701, pageEnd=2713, url=null, language=null, rfNumber=[37], rfOrder=36, authorNames=null, journalName=Chem Sci, refType=null, unstructuredReference=Ragoza M, Masuda T, Koes DR. Generating 3D molecules conditional on receptor binding sites with deep generative models[J]. Chem Sci, 2022, 13: 2701-2713., articleTitle=Generating 3D molecules conditional on receptor binding sites with deep generative models, refAbstract=null), Reference(id=1198960260644963267, tenantId=1146029695717560320, journalId=1189982191388893191, articleId=1198656155770782448, doi=10.1002/minf.201700111, pmid=null, pmcid=null, year=2018, volume=37, issue=null, pageStart=1700111, pageEnd=null, url=null, language=null, rfNumber=[38], rfOrder=37, authorNames=null, journalName=Mol Inform, refType=null, unstructuredReference=Gupta A, Müller AT, Huisman BJH, et al. Generative recurrent networks for de novo drug design[J]. Mol Inform, 2018, 37: 1700111., articleTitle=Generative recurrent networks for de novo drug design, refAbstract=null), Reference(id=1198960260875650000, tenantId=1146029695717560320, journalId=1189982191388893191, articleId=1198656155770782448, doi=10.1186/s13321-020-00446-3, pmid=null, pmcid=null, year=2020, volume=12, issue=null, pageStart=42, pageEnd=null, url=null, language=null, rfNumber=[39], rfOrder=38, authorNames=null, journalName=J Cheminform, refType=null, unstructuredReference=Li XY, Xu YQ, Yao HQ, et al. Chemical space exploration based on recurrent neural networks: applications in discovering kinase inhibitors[J]. J Cheminform, 2020, 12: 42., articleTitle=Chemical space exploration based on recurrent neural networks: applications in discovering kinase inhibitors, refAbstract=null), Reference(id=1198960261068587993, tenantId=1146029695717560320, journalId=1189982191388893191, articleId=1198656155770782448, doi=10.1038/s41598-019-47148-x, pmid=null, pmcid=null, year=2019, volume=9, issue=null, pageStart=10752, pageEnd=null, url=null, language=null, rfNumber=[40], rfOrder=39, authorNames=null, journalName=Sci Rep, refType=null, unstructuredReference=Zhou Z, Kearnes S, Li L, et al. Optimization of molecules via deep reinforcement learning[J]. Sci Rep, 2019, 9: 10752., articleTitle=Optimization of molecules via deep reinforcement learning, refAbstract=null), Reference(id=1198960261181834207, tenantId=1146029695717560320, journalId=1189982191388893191, articleId=1198656155770782448, doi=10.1038/s42256-021-00403-1, pmid=null, pmcid=null, year=2021, volume=3, issue=null, pageStart=914, pageEnd=922, url=null, language=null, rfNumber=[41], rfOrder=40, authorNames=null, journalName=Nat Mach Intell, refType=null, unstructuredReference=Wang J, Hsieh CY, Wang M, et al. Multi-constraint molecular generation based on conditional transformer, knowledge distillation and reinforcement learning[J]. Nat Mach Intell, 2021, 3: 914-922., articleTitle=Multi-constraint molecular generation based on conditional transformer, knowledge distillation and reinforcement learning, refAbstract=null), Reference(id=1198960261395743720, tenantId=1146029695717560320, journalId=1189982191388893191, articleId=1198656155770782448, doi=10.1186/s13321-021-00565-5, pmid=null, pmcid=null, year=2021, volume=13, issue=null, pageStart=87, pageEnd=null, url=null, language=null, rfNumber=[42], rfOrder=41, authorNames=null, journalName=J Cheminform, refType=null, unstructuredReference=Zheng SJ, Lei ZR, Ai HT, et al. Deep scaffold hopping with multimodal transformer neural networks[J]. J Cheminform, 2021, 13: 87., articleTitle=Deep scaffold hopping with multimodal transformer neural networks, refAbstract=null), Reference(id=1198960261630624756, tenantId=1146029695717560320, journalId=1189982191388893191, articleId=1198656155770782448, doi=null, pmid=null, pmcid=null, year=null, volume=null, issue=null, pageStart=null, pageEnd=null, url=null, language=null, rfNumber=[43], rfOrder=42, authorNames=null, journalName=null, refType=null, unstructuredReference=Hoogeboom E, Satorras VG, Vignac C, et al. Equivariant diffusion for molecule generation in 3D [C] // Proceedings of the 39th International Conference on Machine Learning. Maryland: International Machine Learning Society, 2022: 8867-8887., articleTitle=null, refAbstract=null), Reference(id=1198960261890671612, tenantId=1146029695717560320, journalId=1189982191388893191, articleId=1198656155770782448, doi=null, pmid=null, pmcid=null, year=null, volume=null, issue=null, pageStart=null, pageEnd=null, url=null, language=null, rfNumber=44, rfOrder=43, authorNames=null, journalName=null, refType=null, unstructuredReference=Schneuing A, Du YQ, Harris C, et al. Structure-based drug design with equivariant diffusion models [J]. arXiv, 2022: 2210.13695., articleTitle=null, refAbstract=null), Reference(id=1198960262175883271, tenantId=1146029695717560320, journalId=1189982191388893191, articleId=1198656155770782448, doi=null, pmid=null, pmcid=null, year=null, volume=null, issue=null, pageStart=null, pageEnd=null, url=null, language=null, rfNumber=[45], rfOrder=44, authorNames=null, journalName=null, refType=null, unstructuredReference=Doersch C. Tutorial on variational autoencoders [J]. arXiv, 2016: 1606.05908., articleTitle=null, refAbstract=null), Reference(id=1198960262364626958, tenantId=1146029695717560320, journalId=1189982191388893191, articleId=1198656155770782448, doi=10.1002/minf.201700123, pmid=null, pmcid=null, year=2018, volume=37, issue=null, pageStart=1700123, pageEnd=null, url=null, language=null, rfNumber=[46], rfOrder=45, authorNames=null, journalName=Mol Inform, refType=null, unstructuredReference=Blaschke T, Olivecrona M, Engkvist O, et al. Application of generative autoencoder in de novo molecular design[J]. Mol Inform, 2018, 37: 1700123., articleTitle=Application of generative autoencoder in de novo molecular design, refAbstract=null), Reference(id=1198960262612090901, tenantId=1146029695717560320, journalId=1189982191388893191, articleId=1198656155770782448, doi=10.1145/3422622, pmid=null, pmcid=null, year=2020, volume=63, issue=null, pageStart=139, pageEnd=144, url=null, language=null, rfNumber=[47], rfOrder=46, authorNames=null, journalName=Commun ACM, refType=null, unstructuredReference=Goodfellow I, Pouget-Abadie J, Mirza M, et al. Generative adversarial networks[J]. Commun ACM, 2020, 63: 139-144., articleTitle=Generative adversarial networks, refAbstract=null), Reference(id=1198960262922469401, tenantId=1146029695717560320, journalId=1189982191388893191, articleId=1198656155770782448, doi=10.1109/MSP.2017.2765202, pmid=null, pmcid=null, year=2018, volume=35, issue=null, pageStart=53, pageEnd=65, url=null, language=null, rfNumber=[48], rfOrder=47, authorNames=null, journalName=IEEE Signal Process Mag, refType=null, unstructuredReference=Creswell A, White T, Dumoulin V, et al. Generative adversarial networks: an overview[J]. IEEE Signal Process Mag, 2018, 35: 53-65., articleTitle=Generative adversarial networks: an overview, refAbstract=null), Reference(id=1198960263144767531, tenantId=1146029695717560320, journalId=1189982191388893191, articleId=1198656155770782448, doi=10.1146/annurev-control-053018-023825, pmid=null, pmcid=null, year=2019, volume=2, issue=null, pageStart=253, pageEnd=279, url=null, language=null, rfNumber=[49], rfOrder=48, authorNames=null, journalName=Annu Rev Control Robot Auton Syst, refType=null, unstructuredReference=Recht B. A tour of reinforcement learning: the view from continuous control[J]. Annu Rev Control Robot Auton Syst, 2019, 2: 253-279., articleTitle=A tour of reinforcement learning: the view from continuous control, refAbstract=null), Reference(id=1198960263299956788, tenantId=1146029695717560320, journalId=1189982191388893191, articleId=1198656155770782448, doi=10.1038/nchem.1243, pmid=null, pmcid=null, year=2012, volume=4, issue=null, pageStart=90, pageEnd=98, url=null, language=null, rfNumber=[50], rfOrder=49, authorNames=null, journalName=Nat Chem, refType=null, unstructuredReference=Bickerton GR, Paolini GV, Besnard J, et al. Quantifying the chemical beauty of drugs[J]. Nat Chem, 2012, 4: 90-98., articleTitle=Quantifying the chemical beauty of drugs, refAbstract=null), Reference(id=1198960263492894779, tenantId=1146029695717560320, journalId=1189982191388893191, articleId=1198656155770782448, doi=10.1186/1758-2946-1-8, pmid=null, pmcid=null, year=2009, volume=1, issue=null, pageStart=8, pageEnd=null, url=null, language=null, rfNumber=[51], rfOrder=50, authorNames=null, journalName=J Cheminform, refType=null, unstructuredReference=Ertl P, Schuffenhauer A. Estimation of synthetic accessibility score of drug-like molecules based on molecular complexity and fragment contributions[J]. J Cheminform, 2009, 1: 8., articleTitle=Estimation of synthetic accessibility score of drug-like molecules based on molecular complexity and fragment contributions, refAbstract=null), Reference(id=1198960263673249862, tenantId=1146029695717560320, journalId=1189982191388893191, articleId=1198656155770782448, doi=null, pmid=null, pmcid=null, year=null, volume=null, issue=null, pageStart=null, pageEnd=null, url=null, language=null, rfNumber=52, rfOrder=51, authorNames=null, journalName=null, refType=null, unstructuredReference=Graves A. Generating sequences with recurrent neural networks [J]. arXiv, 2013: 1308.0850., articleTitle=null, refAbstract=null), Reference(id=1198960263916519497, tenantId=1146029695717560320, journalId=1189982191388893191, articleId=1198656155770782448, doi=10.1162/neco_a_01199, pmid=null, pmcid=null, year=2019, volume=31, issue=null, pageStart=1235, pageEnd=1270, url=null, language=null, rfNumber=[53], rfOrder=52, authorNames=null, journalName=Neural Comput, refType=null, unstructuredReference=Yu Y, Si X, Hu C, et al. A review of recurrent neural networks: LSTM cells and network architectures[J]. Neural Comput, 2019, 31: 1235-1270., articleTitle=A review of recurrent neural networks: LSTM cells and network architectures, refAbstract=null), Reference(id=1198960264180760658, tenantId=1146029695717560320, journalId=1189982191388893191, articleId=1198656155770782448, doi=10.1126/sciadv.aap7885, pmid=null, pmcid=null, year=2018, volume=4, issue=null, pageStart=eaap7885, pageEnd=null, url=null, language=null, rfNumber=[54], rfOrder=53, authorNames=null, journalName=Sci Adv, refType=null, unstructuredReference=Popova M, Isayev O, Tropsha A. Deep reinforcement learning for de novo drug design[J]. Sci Adv, 2018, 4: eaap7885., articleTitle=Deep reinforcement learning for de novo drug design, refAbstract=null), Reference(id=1198960264482750554, tenantId=1146029695717560320, journalId=1189982191388893191, articleId=1198656155770782448, doi=10.1038/nature16961, pmid=null, pmcid=null, year=2016, volume=529, issue=null, pageStart=484, pageEnd=489, url=null, language=null, rfNumber=[55], rfOrder=54, authorNames=null, journalName=Nature, refType=null, unstructuredReference=Silver D, Huang A, Maddison CJ, et al. Mastering the game of go with deep neural networks and tree search[J]. Nature, 2016, 529: 484-489., articleTitle=Mastering the game of go with deep neural networks and tree search, refAbstract=null), Reference(id=1198960264759574624, tenantId=1146029695717560320, journalId=1189982191388893191, articleId=1198656155770782448, doi=10.1039/D1SC04444C, pmid=null, pmcid=null, year=2021, volume=12, issue=null, pageStart=13664, pageEnd=13675, url=null, language=null, rfNumber=[56], rfOrder=55, authorNames=null, journalName=Chem Sci, refType=null, unstructuredReference=Li Y, Pei J, Lai L. Structure-based de novo drug design using 3D deep generative models[J]. Chem Sci, 2021, 12: 13664-13675., articleTitle=Structure-based de novo drug design using 3D deep generative models, refAbstract=null), Reference(id=1198960265011232876, tenantId=1146029695717560320, journalId=1189982191388893191, articleId=1198656155770782448, doi=10.1021/acsomega.1c05145, pmid=null, pmcid=null, year=2021, volume=6, issue=null, pageStart=33864, pageEnd=33873, url=null, language=null, rfNumber=[57], rfOrder=56, authorNames=null, journalName=ACS Omega, refType=null, unstructuredReference=Yang L, Yang G, Bing Z, et al. Transformer-based generative model accelerating the development of novel BRAF inhibitors[J]. ACS Omega, 2021, 6: 33864-33873., articleTitle=Transformer-based generative model accelerating the development of novel BRAF inhibitors, refAbstract=null), Reference(id=1198960265208365170, tenantId=1146029695717560320, journalId=1189982191388893191, articleId=1198656155770782448, doi=null, pmid=null, pmcid=null, year=null, volume=null, issue=null, pageStart=null, pageEnd=null, url=null, language=null, rfNumber=[58], rfOrder=57, authorNames=null, journalName=null, refType=null, unstructuredReference=Vaswani A, Shazeer N, Parmar N, et al. Attention is all you need [C] // Advances in Neural Information Processing Systems. California: Curran Associates, 2017: 5998–6008., articleTitle=null, refAbstract=null), Reference(id=1198960265371943033, tenantId=1146029695717560320, journalId=1189982191388893191, articleId=1198656155770782448, doi=null, pmid=null, pmcid=null, year=null, volume=null, issue=null, pageStart=null, pageEnd=null, url=null, language=null, rfNumber=59, rfOrder=58, authorNames=null, journalName=null, refType=null, unstructuredReference=Devlin J, Chang MW, Lee K, et al. Bert: pre-training of deep bidirectional transformers for language understanding [J]. arXiv, 2018: 1810.04805., articleTitle=null, refAbstract=null), Reference(id=1198960265543909502, tenantId=1146029695717560320, journalId=1189982191388893191, articleId=1198656155770782448, doi=10.1007/s11431-020-1647-3, pmid=null, pmcid=null, year=2020, volume=63, issue=null, pageStart=1872, pageEnd=1897, url=null, language=null, rfNumber=[60], rfOrder=59, authorNames=null, journalName=Sci China Technol Sci, refType=null, unstructuredReference=Qiu XP, Sun TX, Xu YG, et al. Pre-trained models for natural language processing: a survey[J]. Sci China Technol Sci, 2020, 63: 1872-1897., articleTitle=Pre-trained models for natural language processing: a survey, refAbstract=null), Reference(id=1198960265732653194, tenantId=1146029695717560320, journalId=1189982191388893191, articleId=1198656155770782448, doi=10.1038/s41598-020-79682-4, pmid=null, pmcid=null, year=2021, volume=11, issue=null, pageStart=321, pageEnd=null, url=null, language=null, rfNumber=[61], rfOrder=60, authorNames=null, journalName=Sci Rep, refType=null, unstructuredReference=Grechishnikova D. Transformer neural network for protein-specific de novo drug generation as a machine translation problem[J]. Sci Rep, 2021, 11: 321., articleTitle=Transformer neural network for protein-specific de novo drug generation as a machine translation problem, refAbstract=null), Reference(id=1198960265866870923, tenantId=1146029695717560320, journalId=1189982191388893191, articleId=1198656155770782448, doi=10.1021/acs.jcim.0c00915, pmid=null, pmcid=null, year=2020, volume=60, issue=null, pageStart=5918, pageEnd=5922, url=null, language=null, rfNumber=[62], rfOrder=61, authorNames=null, journalName=J Chem Inf Model, refType=null, unstructuredReference=Blaschke T, Arus-Pous J, Chen H, et al. REINVENT 2.0: an AI tool for de novo drug design[J]. J Chem Inf Model, 2020, 60: 5918-5922., articleTitle=REINVENT 2.0: an AI tool for de novo drug design, refAbstract=null), Reference(id=1198960266017865875, tenantId=1146029695717560320, journalId=1189982191388893191, articleId=1198656155770782448, doi=null, pmid=null, pmcid=null, year=null, volume=null, issue=null, pageStart=null, pageEnd=null, url=null, language=null, rfNumber=63, rfOrder=62, authorNames=null, journalName=null, refType=null, unstructuredReference=Yang L, Zhang ZL, Song Y, et al. Diffusion models: a comprehensive survey of methods and applications [J]. arXiv, 2022: 2209.00796., articleTitle=null, refAbstract=null), Reference(id=1198960266139500697, tenantId=1146029695717560320, journalId=1189982191388893191, articleId=1198656155770782448, doi=null, pmid=null, pmcid=null, year=2022, volume=23, issue=null, pageStart=1, pageEnd=33, url=null, language=null, rfNumber=[64], rfOrder=63, authorNames=null, journalName=J Mach Learn Res, refType=null, unstructuredReference=Ho J, Saharia C, Chan W, et al. Cascaded diffusion models for high fidelity image generation[J]. J Mach Learn Res, 2022, 23: 1-33, articleTitle=Cascaded diffusion models for high fidelity image generation, refAbstract=null), Reference(id=1198960266282107039, tenantId=1146029695717560320, journalId=1189982191388893191, articleId=1198656155770782448, doi=null, pmid=null, pmcid=null, year=null, volume=null, issue=null, pageStart=null, pageEnd=null, url=null, language=null, rfNumber=65, rfOrder=64, authorNames=null, journalName=null, refType=null, unstructuredReference=Li X, Thickstun J, Gulrajani I, et al. Diffusion-lm improves controllable text generation [J]. arXiv, 2022: 2205.14217., articleTitle=null, refAbstract=null), Reference(id=1198960266395353253, tenantId=1146029695717560320, journalId=1189982191388893191, articleId=1198656155770782448, doi=10.1016/j.cell.2014.03.012, pmid=null, pmcid=null, year=2014, volume=157, issue=null, pageStart=255, pageEnd=266, url=null, language=null, rfNumber=[66], rfOrder=65, authorNames=null, journalName=Cell, refType=null, unstructuredReference=Evans RM, Mangelsdorf DJ. Nuclear receptors, RXR, and the big bang[J]. Cell, 2014, 157: 255-266., articleTitle=Nuclear receptors, RXR, and the big bang, refAbstract=null), Reference(id=1198960266512793768, tenantId=1146029695717560320, journalId=1189982191388893191, articleId=1198656155770782448, doi=10.1021/jm500132p, pmid=null, pmcid=null, year=2014, volume=57, issue=null, pageStart=3117, pageEnd=3125, url=null, language=null, rfNumber=[67], rfOrder=66, authorNames=null, journalName=J Med Chem, refType=null, unstructuredReference=Lagarde N, Ben Nasr N, Jeremie A, et al. NRLiSt BDB, the manually curated nuclear receptors ligands and structures benchmarking database[J]. J Med Chem, 2014, 57: 3117-3125., articleTitle=NRLiSt BDB, the manually curated nuclear receptors ligands and structures benchmarking database, refAbstract=null), Reference(id=1198960266672177329, tenantId=1146029695717560320, journalId=1189982191388893191, articleId=1198656155770782448, doi=10.1002/minf.201700153, pmid=null, pmcid=null, year=2018, volume=37, issue=null, pageStart=1700153, pageEnd=null, url=null, language=null, rfNumber=[68], rfOrder=67, authorNames=null, journalName=Mol Inform, refType=null, unstructuredReference=Merk D, Friedrich L, Grisoni F, et al. De novo design of bioactive small molecules by artificial intelligence[J]. Mol Inform, 2018, 37: 1700153., articleTitle=De novo design of bioactive small molecules by artificial intelligence, refAbstract=null), Reference(id=1198960266802200756, tenantId=1146029695717560320, journalId=1189982191388893191, articleId=1198656155770782448, doi=10.1073/pnas.1320001111, pmid=null, pmcid=null, year=2014, volume=111, issue=null, pageStart=4067, pageEnd=4072, url=null, language=null, rfNumber=[69], rfOrder=68, authorNames=null, journalName=Proc Natl Acad Sci, refType=null, unstructuredReference=Reker D, Rodrigues T, Schneider P, et al. Identifying the macromolecular targets of de novo-designed chemical entities through self-organizing map consensus[J]. Proc Natl Acad Sci, 2014, 111: 4067-4072., articleTitle=Identifying the macromolecular targets of de novo-designed chemical entities through self-organizing map consensus, refAbstract=null), Reference(id=1198960266907058363, tenantId=1146029695717560320, journalId=1189982191388893191, articleId=1198656155770782448, doi=10.1016/0920-9964(91)90027-O, pmid=null, pmcid=null, year=1991, volume=4, issue=null, pageStart=81, pageEnd=90, url=null, language=null, rfNumber=[70], rfOrder=69, authorNames=null, journalName=Schizophr Res, refType=null, unstructuredReference=Marder SR, Wirshing WC, Van Putten T. Drug treatment of schizophrenia: overview of recent research[J]. Schizophr Res, 1991, 4: 81-90., articleTitle=Drug treatment of schizophrenia: overview of recent research, refAbstract=null), Reference(id=1198960267045470401, tenantId=1146029695717560320, journalId=1189982191388893191, articleId=1198656155770782448, doi=10.1016/j.ejmech.2020.112572, pmid=null, pmcid=null, year=2020, volume=204, issue=null, pageStart=112572, pageEnd=null, url=null, language=null, rfNumber=[71], rfOrder=70, authorNames=null, journalName=Eur J Med Chem, refType=null, unstructuredReference=Tan XQ, Jiang XR, He Y, et al. Automated design and optimization of multitarget schizophrenia drug candidates by deep learning[J]. Eur J Med Chem, 2020, 204: 112572., articleTitle=Automated design and optimization of multitarget schizophrenia drug candidates by deep learning, refAbstract=null), Reference(id=1198960267171299527, tenantId=1146029695717560320, journalId=1189982191388893191, articleId=1198656155770782448, doi=null, pmid=null, pmcid=null, year=2022, volume=187, issue=null, pageStart=106552, pageEnd=null, url=null, language=null, rfNumber=[72], rfOrder=71, authorNames=null, journalName=Pharmacol Res, refType=null, unstructuredReference=Roskoski Jr R. Properties of FDA-approved small molecule protein kinase inhibitors: a 2023 update[J]. Pharmacol Res, 2022, 187: 106552, articleTitle=Properties of FDA-approved small molecule protein kinase inhibitors: a 2023 update, refAbstract=null), Reference(id=1198960267322294475, tenantId=1146029695717560320, journalId=1189982191388893191, articleId=1198656155770782448, doi=10.1016/j.phrs.2015.10.021, pmid=null, pmcid=null, year=2016, volume=103, issue=null, pageStart=26, pageEnd=48, url=null, language=null, rfNumber=[73], rfOrder=72, authorNames=null, journalName=Pharmacol Res, refType=null, unstructuredReference=Roskoski Jr R. Classification of small molecule protein kinase inhibitors based upon the structures of their drug-enzyme complexes[J]. Pharmacol Res, 2016, 103: 26-48., articleTitle=Classification of small molecule protein kinase inhibitors based upon the structures of their drug-enzyme complexes, refAbstract=null), Reference(id=1198960267477483729, tenantId=1146029695717560320, journalId=1189982191388893191, articleId=1198656155770782448, doi=10.1021/acs.molpharmaceut.7b01137, pmid=null, pmcid=null, year=2018, volume=15, issue=null, pageStart=4386, pageEnd=4397, url=null, language=null, rfNumber=[74], rfOrder=73, authorNames=null, journalName=Mol Pharm, refType=null, unstructuredReference=Putin E, Asadulaev A, Vanhaelen Q, et al. Adversarial threshold neural computer for molecular de novo design[J]. Mol Pharm, 2018, 15: 4386-4397., articleTitle=Adversarial threshold neural computer for molecular de novo design, refAbstract=null), Reference(id=1198960267624284374, tenantId=1146029695717560320, journalId=1189982191388893191, articleId=1198656155770782448, doi=10.1038/s41587-020-0418-2, pmid=null, pmcid=null, year=2020, volume=38, issue=null, pageStart=143, pageEnd=145, url=null, language=null, rfNumber=[75], rfOrder=74, authorNames=null, journalName=Nat Biotechnol, refType=null, unstructuredReference=Walters WP, Murcko M. Assessing the impact of generative AI on medicinal chemistry[J]. Nat Biotechnol, 2020, 38: 143-145., articleTitle=Assessing the impact of generative AI on medicinal chemistry, refAbstract=null), Reference(id=1198960267783667933, tenantId=1146029695717560320, journalId=1189982191388893191, articleId=1198656155770782448, doi=10.1021/acs.jmedchem.1c01205, pmid=null, pmcid=null, year=2022, volume=65, issue=null, pageStart=103, pageEnd=119, url=null, language=null, rfNumber=[76], rfOrder=75, authorNames=null, journalName=J Med Chem, refType=null, unstructuredReference=Tan X, Li C, Yang R, et al. Discovery of pyrazolo[3, 4-d]pyridazinone derivatives as selective DDR1 inhibitors via deep learning based design, synthesis, and biological evaluation[J]. J Med Chem, 2022, 65: 103-119., articleTitle=Discovery of pyrazolo[3, 4-d]pyridazinone derivatives as selective DDR1 inhibitors via deep learning based design, synthesis, and biological evaluation, refAbstract=null), Reference(id=1198960268102435045, tenantId=1146029695717560320, journalId=1189982191388893191, articleId=1198656155770782448, doi=10.1038/s41467-022-34692-w, pmid=null, pmcid=null, year=2022, volume=13, issue=null, pageStart=6891, pageEnd=null, url=null, language=null, rfNumber=[77], rfOrder=76, authorNames=null, journalName=Nat Commun, refType=null, unstructuredReference=Li YS, Zhang LT, Wang YF, et al. Generative deep learning enables the discovery of a potent and selective RIPK1 inhibitor[J]. Nat Commun, 2022, 13: 6891., articleTitle=Generative deep learning enables the discovery of a potent and selective RIPK1 inhibitor, refAbstract=null), Reference(id=1198960268236652778, tenantId=1146029695717560320, journalId=1189982191388893191, articleId=1198656155770782448, doi=10.1021/acs.jmedchem.9b01721, pmid=null, pmcid=null, year=2020, volume=63, issue=null, pageStart=1337, pageEnd=1360, url=null, language=null, rfNumber=[78], rfOrder=77, authorNames=null, journalName=J Med Chem, refType=null, unstructuredReference=Yang YX, Zhang RK, Li ZJ, et al. Discovery of highly potent, selective, and orally efficacious p300/CBP histone acetyltransferases inhibitors[J]. J Med Chem, 2020, 63: 1337-1360., articleTitle=Discovery of highly potent, selective, and orally efficacious p300/CBP histone acetyltransferases inhibitors, refAbstract=null), Reference(id=1198960268354093296, tenantId=1146029695717560320, journalId=1189982191388893191, articleId=1198656155770782448, doi=10.1039/D1CC03392A, pmid=null, pmcid=null, year=2021, volume=57, issue=null, pageStart=10588, pageEnd=10591, url=null, language=null, rfNumber=[79], rfOrder=78, authorNames=null, journalName=Chem Commun, refType=null, unstructuredReference=Wang ZQ, Ran T, Xu F, et al. Deep learning-driven scaffold hopping in the discovery of Akt kinase inhibitors[J]. Chem Commun, 2021, 57: 10588-10591., articleTitle=Deep learning-driven scaffold hopping in the discovery of Akt kinase inhibitors, refAbstract=null), Reference(id=1198960268467339507, tenantId=1146029695717560320, journalId=1189982191388893191, articleId=1198656155770782448, doi=null, pmid=null, pmcid=null, year=2022, volume=247, issue=null, pageStart=115034, pageEnd=null, url=null, language=null, rfNumber=[80], rfOrder=79, authorNames=null, journalName=Eur J Med Chem, refType=null, unstructuredReference=Song SK, Tang HT, Ran T, et al. Application of deep generative model for design of pyrrolo[2, 3-d] pyrimidine derivatives as new selective TANK binding kinase 1 (TBK1) inhibitors[J]. Eur J Med Chem, 2022, 247: 115034, articleTitle=Application of deep generative model for design of pyrrolo[2, 3-d] pyrimidine derivatives as new selective TANK binding kinase 1 (TBK1) inhibitors, refAbstract=null), Reference(id=1198960268584780023, tenantId=1146029695717560320, journalId=1189982191388893191, articleId=1198656155770782448, doi=null, pmid=null, pmcid=null, year=2021, volume=56, issue=null, pageStart=2136, pageEnd=2145, url=http://www.yxxb.com.cn/aps/cn/article/doi/10.16438/j.0513-4870.2021-0052, language=null, rfNumber=[81], rfOrder=80, authorNames=null, journalName=Acta Pharm Sin (药学学报), refType=null, unstructuredReference=Liu RZ, Song JK, Liu AL, et al. Progress on the application of artificial intelligence technology in ligand-based and receptor structure-based drug screening[J]. Acta Pharm Sin (药学学报), 2021, 56: 2136-2145, articleTitle=Progress on the application of artificial intelligence technology in ligand-based and receptor structure-based drug screening, refAbstract=null), Reference(id=1198960268685443322, tenantId=1146029695717560320, journalId=1189982191388893191, articleId=1198656155770782448, doi=10.1186/s13321-019-0341-z, pmid=null, pmcid=null, year=2019, volume=11, issue=null, pageStart=20, pageEnd=null, url=null, language=null, rfNumber=[82], rfOrder=81, authorNames=null, journalName=J Cheminform, refType=null, unstructuredReference=Arús-Pous J, Blaschke T, Ulander S, et al. Exploring the GDB-13 chemical space using deep generative models[J]. J Cheminform, 2019, 11: 20., articleTitle=Exploring the GDB-13 chemical space using deep generative models, refAbstract=null), Reference(id=1198960268815466753, tenantId=1146029695717560320, journalId=1189982191388893191, articleId=1198656155770782448, doi=10.1021/jm030543u, pmid=null, pmcid=null, year=2004, volume=47, issue=null, pageStart=2768, pageEnd=2775, url=null, language=null, rfNumber=[83], rfOrder=82, authorNames=null, journalName=J Med Chem, refType=null, unstructuredReference=Pierce AC, Rao G, Bemis GW. BREED:  generating novel inhibitors through hybridization of known ligands. Application to CDK2, P38, and HIV protease[J]. J Med Chem, 2004, 47: 2768-2775., articleTitle=BREED:  generating novel inhibitors through hybridization of known ligands. Application to CDK2, P38, and HIV protease, refAbstract=null), Reference(id=1198960268953878788, tenantId=1146029695717560320, journalId=1189982191388893191, articleId=1198656155770782448, doi=10.1002/anie.201814681, pmid=null, pmcid=null, year=2019, volume=58, issue=null, pageStart=10792, pageEnd=10803, url=null, language=null, rfNumber=[84], rfOrder=83, authorNames=null, journalName=Angew Chem Int Ed, refType=null, unstructuredReference=Schneider G, Clark DE. Automated de novo drug design: are we nearly there yet?[J]. Angew Chem Int Ed, 2019, 58: 10792-10803., articleTitle=Automated de novo drug design: are we nearly there yet?, refAbstract=null), Reference(id=1198960269050347781, tenantId=1146029695717560320, journalId=1189982191388893191, articleId=1198656155770782448, doi=10.1093/bib/bby061, pmid=null, pmcid=null, year=2019, volume=20, issue=null, pageStart=1878, pageEnd=1912, url=null, language=null, rfNumber=[85], rfOrder=84, authorNames=null, journalName=Brief Bioinform, refType=null, unstructuredReference=Rifaioglu AS, Atas H, Martin MJ, et al. Recent applications of deep learning and machine intelligence on in silico drug discovery: methods, tools and databases[J]. Brief Bioinform, 2019, 20: 1878-1912., articleTitle=Recent applications of deep learning and machine intelligence on in silico drug discovery: methods, tools and databases, refAbstract=null), Reference(id=1198960269209731339, tenantId=1146029695717560320, journalId=1189982191388893191, articleId=1198656155770782448, doi=10.1021/acs.jcim.0c00675, pmid=null, pmcid=null, year=2020, volume=60, issue=null, pageStart=6065, pageEnd=6073, url=null, language=null, rfNumber=[86], rfOrder=85, authorNames=null, journalName=J Chem Inf Model, refType=null, unstructuredReference=Irwin JJ, Tang KG, Young J, et al. ZINC20—a free ultralarge-scale chemical database for ligand discovery[J]. J Chem Inf Model, 2020, 60: 6065-6073., articleTitle=ZINC20—a free ultralarge-scale chemical database for ligand discovery, refAbstract=null), Reference(id=1198960269322977550, tenantId=1146029695717560320, journalId=1189982191388893191, articleId=1198656155770782448, doi=10.1093/nar/gky1075, pmid=null, pmcid=null, year=2019, volume=47, issue=null, pageStart=D930, pageEnd=D940, url=null, language=null, rfNumber=[87], rfOrder=86, authorNames=null, journalName=Nucleic Acids Res, refType=null, unstructuredReference=Mendez D, Gaulton A, Bento AP, et al. ChEMBL: towards direct deposition of bioassay data[J]. Nucleic Acids Res, 2019, 47: D930-D940., articleTitle=ChEMBL: towards direct deposition of bioassay data, refAbstract=null), Reference(id=1198960269436223759, tenantId=1146029695717560320, journalId=1189982191388893191, articleId=1198656155770782448, doi=10.1021/ja902302h, pmid=null, pmcid=null, year=2009, volume=131, issue=null, pageStart=8732, pageEnd=8733, url=null, language=null, rfNumber=[88], rfOrder=87, authorNames=null, journalName=J Am Chem Soc, refType=null, unstructuredReference=Blum LC, Reymond JL. 970 million druglike small molecules for virtual screening in the chemical universe database GDB-13[J]. J Am Chem Soc, 2009, 131: 8732-8733., articleTitle=970 million druglike small molecules for virtual screening in the chemical universe database GDB-13, refAbstract=null), Reference(id=1198960269557858580, tenantId=1146029695717560320, journalId=1189982191388893191, articleId=1198656155770782448, doi=10.1038/s42256-020-0160-y, pmid=null, pmcid=null, year=2020, volume=2, issue=null, pageStart=171, pageEnd=180, url=null, language=null, rfNumber=[89], rfOrder=88, authorNames=null, journalName=Nat Mach Intell, refType=null, unstructuredReference=Moret M, Friedrich L, Grisoni F, et al. Generative molecular design in low data regimes[J]. Nat Mach Intell, 2020, 2: 171-180., articleTitle=Generative molecular design in low data regimes, refAbstract=null), Reference(id=1198960269683687702, tenantId=1146029695717560320, journalId=1189982191388893191, articleId=1198656155770782448, doi=10.1021/acs.jcim.8b00839, pmid=null, pmcid=null, year=2019, volume=59, issue=null, pageStart=1096, pageEnd=1108, url=null, language=null, rfNumber=[90], rfOrder=89, authorNames=null, journalName=J Chem Inf Model, refType=null, unstructuredReference=Brown N, Fiscato M, Segler MHS, et al. GuacaMol: benchmarking models for de novo molecular design[J]. J Chem Inf Model, 2019, 59: 1096-1108., articleTitle=GuacaMol: benchmarking models for de novo molecular design, refAbstract=null), Reference(id=1198960269792739609, tenantId=1146029695717560320, journalId=1189982191388893191, articleId=1198656155770782448, doi=10.1007/s12551-016-0247-1, pmid=null, pmcid=null, year=2017, volume=9, issue=null, pageStart=91, pageEnd=102, url=null, language=null, rfNumber=[91], rfOrder=90, authorNames=null, journalName=Biophys Rev, refType=null, unstructuredReference=Pagadala NS, Syed K, Tuszynski J. Software for molecular docking: a review[J]. Biophys Rev, 2017, 9: 91-102., articleTitle=Software for molecular docking: a review, refAbstract=null), Reference(id=1198960269922763040, tenantId=1146029695717560320, journalId=1189982191388893191, articleId=1198656155770782448, doi=10.1021/acs.jcim.7b00403, pmid=null, pmcid=null, year=2018, volume=58, issue=null, pageStart=916, pageEnd=932, url=null, language=null, rfNumber=[92], rfOrder=91, authorNames=null, journalName=J Chem Inf Model, refType=null, unstructuredReference=Wallach I, Heifets A. Most ligand-based classification benchmarks reward memorization rather than generalization[J]. J Chem Inf Model, 2018, 58: 916-932., articleTitle=Most ligand-based classification benchmarks reward memorization rather than generalization, refAbstract=null), Reference(id=1198960270073757985, tenantId=1146029695717560320, journalId=1189982191388893191, articleId=1198656155770782448, doi=10.1016/j.drudis.2018.01.039, pmid=null, pmcid=null, year=2018, volume=23, issue=null, pageStart=1241, pageEnd=1250, url=null, language=null, rfNumber=[93], rfOrder=92, authorNames=null, journalName=Drug Discov Today, refType=null, unstructuredReference=Chen HM, Engkvist O, Wang YH, et al. The rise of deep learning in drug discovery[J]. Drug Discov Today, 2018, 23: 1241-1250., articleTitle=The rise of deep learning in drug discovery, refAbstract=null)], funds=[Fund(id=1198960254240260517, tenantId=1146029695717560320, journalId=1189982191388893191, articleId=1198656155770782448, awardId=2021-I2M-1-069, language=CN, fundingSource=中国医学科学院医学与健康科技创新工程(2021-I2M-1-069), fundOrder=null, country=null), Fund(id=1198960254449975738, tenantId=1146029695717560320, journalId=1189982191388893191, articleId=1198656155770782448, awardId=2018ZX09711001-012-003, language=CN, fundingSource=国家重大新药创制科技重大专项(2018ZX09711001-012-003), fundOrder=null, country=null)], companyList=[AuthorCompany(id=1198960248963826525, tenantId=1146029695717560320, journalId=1189982191388893191, articleId=1198656155770782448, xref=null, ext=[AuthorCompanyExt(id=1198960248972215135, tenantId=1146029695717560320, journalId=1189982191388893191, articleId=1198656155770782448, companyId=1198960248963826525, language=EN, country=null, province=null, city=null, postcode=null, companyName=null, departmentName=null, remark=State Key Laboratory of Bioactive Substance and Function of Natural Medicines, Institute of Materia Medica, Chinese Academy of Medical Sciences & Peking Union Medical College, Beijing 100050, China), AuthorCompanyExt(id=1198960248976409440, tenantId=1146029695717560320, journalId=1189982191388893191, articleId=1198656155770782448, companyId=1198960248963826525, language=CN, country=null, province=null, city=null, postcode=null, companyName=null, departmentName=null, remark=中国医学科学院、北京协和医学院药物研究所, 天然药物活性物质与功能国家重点实验室, 北京 100050)])], figs=[ArticleFig(id=1198960253225238837, tenantId=1146029695717560320, journalId=1189982191388893191, articleId=1198656155770782448, language=EN, label=null, caption=null, figureFileSmall=lXJF9S+ug6o6k7aEC2vCPw==, figureFileBig=bz8NNQbwH4iPyslSIs73kA==, tableContent=null), ArticleFig(id=1198960253397205324, tenantId=1146029695717560320, journalId=1189982191388893191, articleId=1198656155770782448, language=CN, label=Figure 1, caption= Architectures of generative deep learning models commonly used in <i>de novo</i> drug design. (a) Variational autoencoder; (b) Generative adversarial network; (c) Recurrent neural network; (d) Reinforcement learning; (e) Transformer; (f) Diffusion , figureFileSmall=lXJF9S+ug6o6k7aEC2vCPw==, figureFileBig=bz8NNQbwH4iPyslSIs73kA==, tableContent=null), ArticleFig(id=1198960253569171807, tenantId=1146029695717560320, journalId=1189982191388893191, articleId=1198656155770782448, language=EN, label=null, caption=null, figureFileSmall=null, figureFileBig=null, tableContent=
Year Tool name Tool architecture Molecular representation Database Open-source repository Ref.
2016 ChemVAE VAE SMILES ZINC/GDB https://github.com/aspuru-guzik-group/chemical_vae [30]
2019 GENTRL VAE SMILES, MG (2D) ZINC https://github.com/insilicomedicine/gentrl [16]
2022 JAEGER VAE MG (2D) Novartis in-house assay https://github.com/Novartis/JAEGER [36]
2022 LiGAN VAE 3D Grids CrossDocked https://github.com/mattragoza/liGAN [37]
2017 ORGAN GAN SMILES GDB https://github.com/gablg1/ORGAN [31]
2018 Segler et al. RNN SMILES ChEMBL https://github.com/jaechanglim/molecule-generator [32]
2018 LSTMChem RNN SMILES ChEMBL https://github.com/topazape/LSTM_Chem [38]
2020 Li et al. RNN SMILES ChEMBL https://github.com/Xyqii/RNN_generator [39]
2019 MolDQN RL MG (2D) not used https://github.com/google-research/google-research/tree/master/mol_dqn [40]
2020 SyntaLinker Transformer SMILES ChEMBL https://github.com/YuYaoYang2333/SyntaLinker [34]
2021 MCMG Transformer SMILES ChEMBL https://github.com/jkwang93/MCMG [41]
2021 DeepHop Transformer SMILES, MG (3D) ChEMBL https://github.com/prokia/deepHops [42]
2022 E(3) EDM Diffusion MG (3D) GDB https://github.com/ehoogeboom/e3_diffusion_for_molecules [43]
2022 DiffSBDD Diffusion MG (3D) CrossDocked/Binding MOAD https://github.com/arneschneuing/DiffSBDD [44]
), ArticleFig(id=1198960253703389552, tenantId=1146029695717560320, journalId=1189982191388893191, articleId=1198656155770782448, language=CN, label=Table 1, caption=

Representative open-source tools for deep molecular generation

, figureFileSmall=null, figureFileBig=null, tableContent=
Year Tool name Tool architecture Molecular representation Database Open-source repository Ref.
2016 ChemVAE VAE SMILES ZINC/GDB https://github.com/aspuru-guzik-group/chemical_vae [30]
2019 GENTRL VAE SMILES, MG (2D) ZINC https://github.com/insilicomedicine/gentrl [16]
2022 JAEGER VAE MG (2D) Novartis in-house assay https://github.com/Novartis/JAEGER [36]
2022 LiGAN VAE 3D Grids CrossDocked https://github.com/mattragoza/liGAN [37]
2017 ORGAN GAN SMILES GDB https://github.com/gablg1/ORGAN [31]
2018 Segler et al. RNN SMILES ChEMBL https://github.com/jaechanglim/molecule-generator [32]
2018 LSTMChem RNN SMILES ChEMBL https://github.com/topazape/LSTM_Chem [38]
2020 Li et al. RNN SMILES ChEMBL https://github.com/Xyqii/RNN_generator [39]
2019 MolDQN RL MG (2D) not used https://github.com/google-research/google-research/tree/master/mol_dqn [40]
2020 SyntaLinker Transformer SMILES ChEMBL https://github.com/YuYaoYang2333/SyntaLinker [34]
2021 MCMG Transformer SMILES ChEMBL https://github.com/jkwang93/MCMG [41]
2021 DeepHop Transformer SMILES, MG (3D) ChEMBL https://github.com/prokia/deepHops [42]
2022 E(3) EDM Diffusion MG (3D) GDB https://github.com/ehoogeboom/e3_diffusion_for_molecules [43]
2022 DiffSBDD Diffusion MG (3D) CrossDocked/Binding MOAD https://github.com/arneschneuing/DiffSBDD [44]
), ArticleFig(id=1198960253837607296, tenantId=1146029695717560320, journalId=1189982191388893191, articleId=1198656155770782448, language=EN, label=null, caption=null, figureFileSmall=null, figureFileBig=null, tableContent=
Year Target/Indication Tool architecture Number of generated molecules Number of validated molecules Chemical structure of representative lead Bioassay (in vitro) Bioassay (in vivo) Ref.
2018 RXRs, PPARs RNN 1 000 5 EC50: 0.06 μmol·L-1 (RXRγ), 2.3 μmol·L-1 (PPARγ) No [68]
2018 Multi-kinases GAN 30 000 50 Inhibition @ 10 μmol·L-1: 101% (SGK1) No [74]
2019 DDR1 VAE 30 000 4 IC50: 10 nmol·L-1 No [16]
2020 p300, CBP RNN 672 1 IC50: 1.8 nmol·L-1 (p300) Yes [78]
2020 CDK4, Pim1 RNN Undisclosed 3 Inhibition @ 10 μmol·L-1: 99.64% (Pim1) No [39]
2020 Schizophrenia RNN 10 000 1 IC50: 216 nmol·L-1 (D2R), 1.64 nmol·L-1 (5-HT2AR), 0.51 nmol·L-1 (5-HT1AR) Yes [71]
2021 Akt Transformer 6 617 1 IC50: 88 nmol·L-1 (Akt1) No [79]
2022 DDR1 RNN 19 929 2 IC50: 10.6 nmol·L-1 Yes [76]
2022 Malaria VAE 282 2 IC50: 1.276 nmol·L-1 (PI4K) No [36]
2022 RIPK1 RNN 79 323 8 IC50: 5.9 nmol·L-1 Yes [77]
2022 TBK1 Transformer 1 101 1 IC50: 22.4 nmol·L-1 No [80]
), ArticleFig(id=1198960253984407947, tenantId=1146029695717560320, journalId=1189982191388893191, articleId=1198656155770782448, language=CN, label=Table 2, caption=

Case studies on leads discovery with AI-based de novo drug design

, figureFileSmall=null, figureFileBig=null, tableContent=
Year Target/Indication Tool architecture Number of generated molecules Number of validated molecules Chemical structure of representative lead Bioassay (in vitro) Bioassay (in vivo) Ref.
2018 RXRs, PPARs RNN 1 000 5 EC50: 0.06 μmol·L-1 (RXRγ), 2.3 μmol·L-1 (PPARγ) No [68]
2018 Multi-kinases GAN 30 000 50 Inhibition @ 10 μmol·L-1: 101% (SGK1) No [74]
2019 DDR1 VAE 30 000 4 IC50: 10 nmol·L-1 No [16]
2020 p300, CBP RNN 672 1 IC50: 1.8 nmol·L-1 (p300) Yes [78]
2020 CDK4, Pim1 RNN Undisclosed 3 Inhibition @ 10 μmol·L-1: 99.64% (Pim1) No [39]
2020 Schizophrenia RNN 10 000 1 IC50: 216 nmol·L-1 (D2R), 1.64 nmol·L-1 (5-HT2AR), 0.51 nmol·L-1 (5-HT1AR) Yes [71]
2021 Akt Transformer 6 617 1 IC50: 88 nmol·L-1 (Akt1) No [79]
2022 DDR1 RNN 19 929 2 IC50: 10.6 nmol·L-1 Yes [76]
2022 Malaria VAE 282 2 IC50: 1.276 nmol·L-1 (PI4K) No [36]
2022 RIPK1 RNN 79 323 8 IC50: 5.9 nmol·L-1 Yes [77]
2022 TBK1 Transformer 1 101 1 IC50: 22.4 nmol·L-1 No [80]
)], attaches=null, journal=Journal(id=1189982048455397383, delFlag=0, nameCn=药学学报, nameEn=Acta Pharmaceutica Sinica, nameHistory1=null, nameHistory2=null, issn=0513-4870, eissn=null, cn=11-2163/R, coden=null, periodic=0, language=CN, oaType=null, ccby=null, superviseOffice=null, ownerOffice=null, pubOffice=null, editorOffice=null, officeType=null, aims=null, clcCode=null, officeProv=null, officeCity=null, officeAddr=null, officeZip=null, officeEmail=null, officePhone=null, editDirector=null, officeDirector=null, officeDirectorPhone=null, officeStaffNum=null, officeEmpNum=null, coverPicUrl=BTxjudbJDVO4PqdBR6On6Q==, journalPrice=null, startedYear=null, abbrevIsoEn=null, journalRemark=null, publicationField=null, createdTime=1761643429151, updatedTime=1788948913902, createdBy=18614031015, updatedBy=13041195026, firstLetterCn=Y, firstLetterEn=Y, subjectCode=Medical and Pharmaceutical Sciences, subjectName=Life Sciences, subjectCodeEn=Medical and Pharmaceutical Sciences, subjectNameEn=null, picCn=BTxjudbJDVO4PqdBR6On6Q==, picEn=c4l1ckL55nWbhl1KrFdWIA==, jcr=null, cjcr=null, exts=[JournalExt(id=1304509553505227679, language=CN, name=药学学报, nameHistory1=null, nameHistory2=null, managedBy=中国科学技术协会, sponsoredBy=中国药学会、中国医学科学院药物研究所, publishedBy=《药学学报》编辑委员会编辑出版, editorOffice=, officeProv=null, officeCity=null, officeAddr=, officeZip=, editDirector=, officeDirector=null, officePhone=null, coverPicUrl=null, journalRemark=, submitArticleUrl=null, websiteUrl=, createdTime=1788948914171, updatedTime=1788948914171, createdBy=13041195026, updatedBy=13041195026, submissionGuidelinesUrl=, submissionAuthorUrl=https://www.yxxb.com.cn/journalx_yxxb/authorLogOn.action, submissionEditorUrl=https://www.yxxb.com.cn/journalx_yxxb/editorLogOn.action, submissionReviewUrl=https://www.yxxb.com.cn/journalx_yxxb/expertLogOn.action, submissionCeEditorUrl=, submissionAeEditorUrl=, option={"copyright":""}), JournalExt(id=1304509553559753632, language=EN, name=Acta Pharmaceutica Sinica, nameHistory1=null, nameHistory2=null, managedBy=, sponsoredBy=, publishedBy=, editorOffice=, officeProv=null, officeCity=null, officeAddr=, officeZip=, editDirector=, officeDirector=null, officePhone=null, coverPicUrl=null, journalRemark=, submitArticleUrl=null, websiteUrl=, createdTime=1788948914184, updatedTime=1788948914184, createdBy=13041195026, updatedBy=13041195026, submissionGuidelinesUrl=, submissionAuthorUrl=https://www.yxxb.com.cn/journalx_yxxb/authorLogOn.action, submissionEditorUrl=https://www.yxxb.com.cn/journalx_yxxb/editorLogOn.action, submissionReviewUrl=https://www.yxxb.com.cn/journalx_yxxb/expertLogOn.action, submissionCeEditorUrl=, submissionAeEditorUrl=, option={"copyright":""})], databaseList=null, tenantJournalId=1189982191388893191, websiteList=[Website(id=1189982271588340489, webName=null, webTitle=null, webDomain=null, webCopyrigh=null, webIpcNo=null, seoTitle=null, seoKeywords=null, seoDescription=null, tenantJournalId=null, journalId=1189982191388893191, journalNameCn=null, journalNameEn=null, grayFlag=null, tenantId=1146029695717560320, platformId=null, journalGroupId=null, journalGroupNameCn=null, journalGroupNameEn=null, type=1, domain=https://castjournals.cast.org.cn/joweb/yxxb/CN, language=CN, createTime=1761643482348, createBy=18614031015, updateTime=1761643498101, updateBy=18614031015, name=药学学报-中文, tplId=1146099689490845704, title=药学学报, delFlag=0, indexPage=/home, props=[WebsiteProps(id=1189982873114448678, tenantId=1146029695717560320, journalId=null, journalGroupId=null, siteId=1189982271588340489, code=articleTextType, value=kx, createTime=1761643625763, updateTime=1761643625763, creator=18614031015, updator=18614031015), WebsiteProps(id=1189982873093477155, tenantId=1146029695717560320, journalId=null, journalGroupId=null, siteId=1189982271588340489, code=banner, value=null, createTime=1761643625758, updateTime=1761643625758, creator=18614031015, updator=18614031015), WebsiteProps(id=1189982873135420201, tenantId=1146029695717560320, journalId=null, journalGroupId=null, siteId=1189982271588340489, code=grayFlag, value=0, createTime=1761643625768, updateTime=1761643625768, creator=18614031015, updator=18614031015), WebsiteProps(id=1189982873085088546, tenantId=1146029695717560320, journalId=null, journalGroupId=null, siteId=1189982271588340489, code=logo, value=https://castjournals.cast.org.cn/joweb/yxxb/CN/file/pic?fileId=w+t2v8bJnX5lh3+hRRJcDA==, createTime=1761643625756, updateTime=1761643625756, creator=18614031015, updator=18614031015), WebsiteProps(id=1189982873152197419, tenantId=1146029695717560320, journalId=null, journalGroupId=null, siteId=1189982271588340489, code=minRunFlag, value=0, createTime=1761643625772, updateTime=1761643625772, creator=18614031015, updator=18614031015), WebsiteProps(id=1189982873110254373, tenantId=1146029695717560320, journalId=null, journalGroupId=null, siteId=1189982271588340489, code=picServerUrl, value=https://castjournals.cast.org.cn/joweb/yxxb/CN/file/pic, createTime=1761643625762, updateTime=1761643625762, creator=18614031015, updator=18614031015), WebsiteProps(id=1189982873143808810, tenantId=1146029695717560320, journalId=null, journalGroupId=null, siteId=1189982271588340489, code=silenceFlag, value=0, createTime=1761643625770, updateTime=1761643625770, creator=18614031015, updator=18614031015), WebsiteProps(id=1189982873101865764, tenantId=1146029695717560320, journalId=null, journalGroupId=null, siteId=1189982271588340489, code=staticResourcePath, value=https://castjournals.cast.org.cn/joweb/cast_kjdb_cn_619/, createTime=1761643625760, updateTime=1761643625760, creator=18614031015, updator=18614031015), WebsiteProps(id=1189982873122837287, tenantId=1146029695717560320, journalId=null, journalGroupId=null, siteId=1189982271588340489, code=themeColor, value=null, createTime=1761643625765, updateTime=1761643625765, creator=18614031015, updator=18614031015), WebsiteProps(id=1189982873127031592, tenantId=1146029695717560320, journalId=null, journalGroupId=null, siteId=1189982271588340489, code=themeStyle, value=null, createTime=1761643625766, updateTime=1761643625766, creator=18614031015, updator=18614031015)]), Website(id=1189982271655449355, webName=null, webTitle=null, webDomain=null, webCopyrigh=null, webIpcNo=null, seoTitle=null, seoKeywords=null, seoDescription=null, tenantJournalId=null, journalId=1189982191388893191, journalNameCn=null, journalNameEn=null, grayFlag=null, tenantId=1146029695717560320, platformId=null, journalGroupId=null, journalGroupNameCn=null, journalGroupNameEn=null, type=1, domain=https://castjournals.cast.org.cn/joweb/yxxb/EN, language=EN, createTime=1761643482364, createBy=18614031015, updateTime=1761643514085, updateBy=18614031015, name=药学学报-英文, tplId=1146101810881728533, title=Acta Pharmaceutica Sinica, delFlag=0, indexPage=/home, props=[WebsiteProps(id=1189982903015633534, tenantId=1146029695717560320, journalId=null, journalGroupId=null, siteId=1189982271655449355, code=articleTextType, value=kx, createTime=1761643632892, updateTime=1761643632892, creator=18614031015, updator=18614031015), WebsiteProps(id=1189982902990467707, tenantId=1146029695717560320, journalId=null, journalGroupId=null, siteId=1189982271655449355, code=banner, value=null, createTime=1761643632886, updateTime=1761643632886, creator=18614031015, updator=18614031015), WebsiteProps(id=1189982903036605057, tenantId=1146029695717560320, journalId=null, journalGroupId=null, siteId=1189982271655449355, code=grayFlag, value=0, createTime=1761643632897, updateTime=1761643632897, creator=18614031015, updator=18614031015), WebsiteProps(id=1189982902982079098, tenantId=1146029695717560320, journalId=null, journalGroupId=null, siteId=1189982271655449355, code=logo, value=https://castjournals.cast.org.cn/joweb/yxxb/EN/file/pic?fileId=w+t2v8bJnX5lh3+hRRJcDA==, createTime=1761643632884, updateTime=1761643632884, creator=18614031015, updator=18614031015), WebsiteProps(id=1189982903053382275, tenantId=1146029695717560320, journalId=null, journalGroupId=null, siteId=1189982271655449355, code=minRunFlag, value=0, createTime=1761643632901, updateTime=1761643632901, creator=18614031015, updator=18614031015), WebsiteProps(id=1189982903007244925, tenantId=1146029695717560320, journalId=null, journalGroupId=null, siteId=1189982271655449355, code=picServerUrl, value=https://castjournals.cast.org.cn/joweb/yxxb/EN/file/pic, createTime=1761643632890, updateTime=1761643632890, creator=18614031015, updator=18614031015), WebsiteProps(id=1189982903044993666, tenantId=1146029695717560320, journalId=null, journalGroupId=null, siteId=1189982271655449355, code=silenceFlag, value=0, createTime=1761643632899, updateTime=1761643632899, creator=18614031015, updator=18614031015), WebsiteProps(id=1189982902998856316, tenantId=1146029695717560320, journalId=null, journalGroupId=null, siteId=1189982271655449355, code=staticResourcePath, value=https://castjournals.cast.org.cn/joweb/cast_kjdb_en_623/, createTime=1761643632888, updateTime=1761643632888, creator=18614031015, updator=18614031015), WebsiteProps(id=1189982903019827839, tenantId=1146029695717560320, journalId=null, journalGroupId=null, siteId=1189982271655449355, code=themeColor, value=null, createTime=1761643632893, updateTime=1761643632893, creator=18614031015, updator=18614031015), WebsiteProps(id=1189982903028216448, tenantId=1146029695717560320, journalId=null, journalGroupId=null, siteId=1189982271655449355, code=themeStyle, value=null, createTime=1761643632895, updateTime=1761643632895, creator=18614031015, updator=18614031015)])], journalTitle=药学学报, weixinUrl=null, journalUrl=https://www.yxxb.com.cn/aps, iacademicId=null, status=1, seqNo=null, journalTitleEn=Acta Pharmaceutica Sinica, journalPhotoCn=BTxjudbJDVO4PqdBR6On6Q==, journalPhotoEn=c4l1ckL55nWbhl1KrFdWIA==, journalFirstLetter=Y, journalRecommend=null, journalNew=null, journalCollection=null, jcrJf=null, cjcrJf=null, jcrJfStr=null, cjcrJfStr=null, submissionFirstDecision=null, sciSubjectClassification=null, casSubjectClassification=null, citeScore=null, totalCitationFrequency=null, icpCode=null, psCode=null, advertisingLicenseCode=null, copyrightInformation=null, country=null, option=, provinceCode=null, provinceName=null, collectFlag=false, interPubPlatform=, interPubPlatformUrl=null), detailUrlCn=https://castjournals.cast.org.cn/joweb/yxxb/CN/10.16438/j.0513-4870.2022-1453, detailUrlEn=https://castjournals.cast.org.cn/joweb/yxxb/EN/10.16438/j.0513-4870.2022-1453, pdfUrlCn=https://castjournals.cast.org.cn/joweb/yxxb/CN/PDF/10.16438/j.0513-4870.2022-1453, pdfUrlEn=https://castjournals.cast.org.cn/joweb/yxxb/EN/PDF/10.16438/j.0513-4870.2022-1453, aliStartDate=null, aliEndDate=null, collectionFlag=false, citedCount=null, citedUrl=null, previewStatus=0, delFlag=0, hasFullText=1, orderTime=1694448000000, fullTextJson=null, articleText=null, reference=null)
收藏切换
基于深度学习的全新药物分子设计: 原理、工具与实践
收藏切换
PDF下载
沈涛 , 王冬梅 , 吴松 , 蒋建东 , 夏杰 *
药学学报 | 综述 2023,58(9): 2610-2622
收起
收藏切换
药学学报 |综述 2023 , 58 (9) : 2610 -2622
基于深度学习的全新药物分子设计: 原理、工具与实践
全屏
[Author(id=1198960249093849965, tenantId=1146029695717560320, journalId=1189982191388893191, articleId=1198656155770782448, orderNo=0, firstName=null, middleName=null, lastName=null, nameCn=null, orcid=null, stid=null, country=null, authorPic=null, dead=0, email=null, emailSecond=null, emailThird=null, correspondingAuthor=0, authorType=1, ext={EN=AuthorExt(id=1198960249228067705, tenantId=1146029695717560320, journalId=1189982191388893191, articleId=1198656155770782448, authorId=1198960249093849965, language=EN, stringName=Tao SHEN, firstName=Tao, middleName=null, lastName=SHEN, prefix=null, suffix=null, authorComment=null, nameInitials=null, affiliation=null, department=null, xref=null, address=State Key Laboratory of Bioactive Substance and Function of Natural Medicines, Institute of Materia Medica, Chinese Academy of Medical Sciences & Peking Union Medical College, Beijing 100050, China, bio=null, bioImg=null, bioContent=null, aboutCorrespAuthor=null), CN=AuthorExt(id=1198960249387451274, tenantId=1146029695717560320, journalId=1189982191388893191, articleId=1198656155770782448, authorId=1198960249093849965, language=CN, stringName=沈涛, firstName=涛, middleName=null, lastName=沈, prefix=null, suffix=null, authorComment=null, nameInitials=null, affiliation=null, department=null, xref=null, address=中国医学科学院、北京协和医学院药物研究所, 天然药物活性物质与功能国家重点实验室, 北京 100050, bio=null, bioImg=null, bioContent=null, aboutCorrespAuthor=null)}, companyList=[AuthorCompany(id=1198960248963826525, tenantId=1146029695717560320, journalId=1189982191388893191, articleId=1198656155770782448, xref=null, ext=[AuthorCompanyExt(id=1198960248972215135, tenantId=1146029695717560320, journalId=1189982191388893191, articleId=1198656155770782448, companyId=1198960248963826525, language=EN, country=null, province=null, city=null, postcode=null, companyName=null, departmentName=null, remark=State Key Laboratory of Bioactive Substance and Function of Natural Medicines, Institute of Materia Medica, Chinese Academy of Medical Sciences & Peking Union Medical College, Beijing 100050, China), AuthorCompanyExt(id=1198960248976409440, tenantId=1146029695717560320, journalId=1189982191388893191, articleId=1198656155770782448, companyId=1198960248963826525, language=CN, country=null, province=null, city=null, postcode=null, companyName=null, departmentName=null, remark=中国医学科学院、北京协和医学院药物研究所, 天然药物活性物质与功能国家重点实验室, 北京 100050)])]), Author(id=1198960249567806369, tenantId=1146029695717560320, journalId=1189982191388893191, articleId=1198656155770782448, orderNo=1, firstName=null, middleName=null, lastName=null, nameCn=null, orcid=null, stid=null, country=null, authorPic=null, dead=0, email=null, emailSecond=null, emailThird=null, correspondingAuthor=0, authorType=1, ext={EN=AuthorExt(id=1198960249731384244, tenantId=1146029695717560320, journalId=1189982191388893191, articleId=1198656155770782448, authorId=1198960249567806369, language=EN, stringName=Dong-mei WANG, firstName=Dong-mei, middleName=null, lastName=WANG, prefix=null, suffix=null, authorComment=null, nameInitials=null, affiliation=null, department=null, xref=null, address=State Key Laboratory of Bioactive Substance and Function of Natural Medicines, Institute of Materia Medica, Chinese Academy of Medical Sciences & Peking Union Medical College, Beijing 100050, China, bio=null, bioImg=null, bioContent=null, aboutCorrespAuthor=null), CN=AuthorExt(id=1198960249873990599, tenantId=1146029695717560320, journalId=1189982191388893191, articleId=1198656155770782448, authorId=1198960249567806369, language=CN, stringName=王冬梅, firstName=冬梅, middleName=null, lastName=王, prefix=null, suffix=null, authorComment=null, nameInitials=null, affiliation=null, department=null, xref=null, address=中国医学科学院、北京协和医学院药物研究所, 天然药物活性物质与功能国家重点实验室, 北京 100050, bio=null, bioImg=null, bioContent=null, aboutCorrespAuthor=null)}, companyList=[AuthorCompany(id=1198960248963826525, tenantId=1146029695717560320, journalId=1189982191388893191, articleId=1198656155770782448, xref=null, ext=[AuthorCompanyExt(id=1198960248972215135, tenantId=1146029695717560320, journalId=1189982191388893191, articleId=1198656155770782448, companyId=1198960248963826525, language=EN, country=null, province=null, city=null, postcode=null, companyName=null, departmentName=null, remark=State Key Laboratory of Bioactive Substance and Function of Natural Medicines, Institute of Materia Medica, Chinese Academy of Medical Sciences & Peking Union Medical College, Beijing 100050, China), AuthorCompanyExt(id=1198960248976409440, tenantId=1146029695717560320, journalId=1189982191388893191, articleId=1198656155770782448, companyId=1198960248963826525, language=CN, country=null, province=null, city=null, postcode=null, companyName=null, departmentName=null, remark=中国医学科学院、北京协和医学院药物研究所, 天然药物活性物质与功能国家重点实验室, 北京 100050)])]), Author(id=1198960249991431126, tenantId=1146029695717560320, journalId=1189982191388893191, articleId=1198656155770782448, orderNo=2, firstName=null, middleName=null, lastName=null, nameCn=null, orcid=null, stid=null, country=null, authorPic=null, dead=0, email=null, emailSecond=null, emailThird=null, correspondingAuthor=0, authorType=1, ext={EN=AuthorExt(id=1198960250121454570, tenantId=1146029695717560320, journalId=1189982191388893191, articleId=1198656155770782448, authorId=1198960249991431126, language=EN, stringName=Song WU, firstName=Song, middleName=null, lastName=WU, prefix=null, suffix=null, authorComment=null, nameInitials=null, affiliation=null, department=null, xref=null, address=State Key Laboratory of Bioactive Substance and Function of Natural Medicines, Institute of Materia Medica, Chinese Academy of Medical Sciences & Peking Union Medical College, Beijing 100050, China, bio=null, bioImg=null, bioContent=null, aboutCorrespAuthor=null), CN=AuthorExt(id=1198960250264060921, tenantId=1146029695717560320, journalId=1189982191388893191, articleId=1198656155770782448, authorId=1198960249991431126, language=CN, stringName=吴松, firstName=松, middleName=null, lastName=吴, prefix=null, suffix=null, authorComment=null, nameInitials=null, affiliation=null, department=null, xref=null, address=中国医学科学院、北京协和医学院药物研究所, 天然药物活性物质与功能国家重点实验室, 北京 100050, bio=null, bioImg=null, bioContent=null, aboutCorrespAuthor=null)}, companyList=[AuthorCompany(id=1198960248963826525, tenantId=1146029695717560320, journalId=1189982191388893191, articleId=1198656155770782448, xref=null, ext=[AuthorCompanyExt(id=1198960248972215135, tenantId=1146029695717560320, journalId=1189982191388893191, articleId=1198656155770782448, companyId=1198960248963826525, language=EN, country=null, province=null, city=null, postcode=null, companyName=null, departmentName=null, remark=State Key Laboratory of Bioactive Substance and Function of Natural Medicines, Institute of Materia Medica, Chinese Academy of Medical Sciences & Peking Union Medical College, Beijing 100050, China), AuthorCompanyExt(id=1198960248976409440, tenantId=1146029695717560320, journalId=1189982191388893191, articleId=1198656155770782448, companyId=1198960248963826525, language=CN, country=null, province=null, city=null, postcode=null, companyName=null, departmentName=null, remark=中国医学科学院、北京协和医学院药物研究所, 天然药物活性物质与功能国家重点实验室, 北京 100050)])]), Author(id=1198960250448609298, tenantId=1146029695717560320, journalId=1189982191388893191, articleId=1198656155770782448, orderNo=3, firstName=null, middleName=null, lastName=null, nameCn=null, orcid=null, stid=null, country=null, authorPic=null, dead=0, email=null, emailSecond=null, emailThird=null, correspondingAuthor=0, authorType=1, ext={EN=AuthorExt(id=1198960250616381478, tenantId=1146029695717560320, journalId=1189982191388893191, articleId=1198656155770782448, authorId=1198960250448609298, language=EN, stringName=Jian-dong JIANG, firstName=Jian-dong, middleName=null, lastName=JIANG, prefix=null, suffix=null, authorComment=null, nameInitials=null, affiliation=null, department=null, xref=null, address=State Key Laboratory of Bioactive Substance and Function of Natural Medicines, Institute of Materia Medica, Chinese Academy of Medical Sciences & Peking Union Medical College, Beijing 100050, China, bio=null, bioImg=null, bioContent=null, aboutCorrespAuthor=null), CN=AuthorExt(id=1198960250775765050, tenantId=1146029695717560320, journalId=1189982191388893191, articleId=1198656155770782448, authorId=1198960250448609298, language=CN, stringName=蒋建东, firstName=建东, middleName=null, lastName=蒋, prefix=null, suffix=null, authorComment=null, nameInitials=null, affiliation=null, department=null, xref=null, address=中国医学科学院、北京协和医学院药物研究所, 天然药物活性物质与功能国家重点实验室, 北京 100050, bio=null, bioImg=null, bioContent=null, aboutCorrespAuthor=null)}, companyList=[AuthorCompany(id=1198960248963826525, tenantId=1146029695717560320, journalId=1189982191388893191, articleId=1198656155770782448, xref=null, ext=[AuthorCompanyExt(id=1198960248972215135, tenantId=1146029695717560320, journalId=1189982191388893191, articleId=1198656155770782448, companyId=1198960248963826525, language=EN, country=null, province=null, city=null, postcode=null, companyName=null, departmentName=null, remark=State Key Laboratory of Bioactive Substance and Function of Natural Medicines, Institute of Materia Medica, Chinese Academy of Medical Sciences & Peking Union Medical College, Beijing 100050, China), AuthorCompanyExt(id=1198960248976409440, tenantId=1146029695717560320, journalId=1189982191388893191, articleId=1198656155770782448, companyId=1198960248963826525, language=CN, country=null, province=null, city=null, postcode=null, companyName=null, departmentName=null, remark=中国医学科学院、北京协和医学院药物研究所, 天然药物活性物质与功能国家重点实验室, 北京 100050)])]), Author(id=1198960250889011268, tenantId=1146029695717560320, journalId=1189982191388893191, articleId=1198656155770782448, orderNo=4, firstName=null, middleName=null, lastName=null, nameCn=null, orcid=null, stid=null, country=null, authorPic=null, dead=0, email=jie.william.xia@hotmail.com, emailSecond=null, emailThird=null, correspondingAuthor=1, authorType=1, ext={EN=AuthorExt(id=1198960251002257487, tenantId=1146029695717560320, journalId=1189982191388893191, articleId=1198656155770782448, authorId=1198960250889011268, language=EN, stringName=Jie XIA, firstName=Jie, middleName=null, lastName=XIA, prefix=null, suffix=null, authorComment=null, nameInitials=null, affiliation=null, department=null, xref=*, address=State Key Laboratory of Bioactive Substance and Function of Natural Medicines, Institute of Materia Medica, Chinese Academy of Medical Sciences & Peking Union Medical College, Beijing 100050, China, bio=null, bioImg=null, bioContent=null, aboutCorrespAuthor=null), CN=AuthorExt(id=1198960251098726491, tenantId=1146029695717560320, journalId=1189982191388893191, articleId=1198656155770782448, authorId=1198960250889011268, language=CN, stringName=夏杰, firstName=杰, middleName=null, lastName=夏, prefix=null, suffix=null, authorComment=null, nameInitials=null, affiliation=null, department=null, xref=*, address=中国医学科学院、北京协和医学院药物研究所, 天然药物活性物质与功能国家重点实验室, 北京 100050, bio=null, bioImg=null, bioContent=null, aboutCorrespAuthor=null)}, companyList=[AuthorCompany(id=1198960248963826525, tenantId=1146029695717560320, journalId=1189982191388893191, articleId=1198656155770782448, xref=null, ext=[AuthorCompanyExt(id=1198960248972215135, tenantId=1146029695717560320, journalId=1189982191388893191, articleId=1198656155770782448, companyId=1198960248963826525, language=EN, country=null, province=null, city=null, postcode=null, companyName=null, departmentName=null, remark=State Key Laboratory of Bioactive Substance and Function of Natural Medicines, Institute of Materia Medica, Chinese Academy of Medical Sciences & Peking Union Medical College, Beijing 100050, China), AuthorCompanyExt(id=1198960248976409440, tenantId=1146029695717560320, journalId=1189982191388893191, articleId=1198656155770782448, companyId=1198960248963826525, language=CN, country=null, province=null, city=null, postcode=null, companyName=null, departmentName=null, remark=中国医学科学院、北京协和医学院药物研究所, 天然药物活性物质与功能国家重点实验室, 北京 100050)])])]
沈涛, 王冬梅, 吴松, 蒋建东, 夏杰*
作者信息
  • 中国医学科学院、北京协和医学院药物研究所, 天然药物活性物质与功能国家重点实验室, 北京 100050
通讯作者:
Deep learning-based de novo drug design: principles, tools and practice
Tao SHEN, Dong-mei WANG, Song WU, Jian-dong JIANG, Jie XIA*
Affiliations
  • State Key Laboratory of Bioactive Substance and Function of Natural Medicines, Institute of Materia Medica, Chinese Academy of Medical Sciences & Peking Union Medical College, Beijing 100050, China
出版时间: 2023-09-12 doi: 10.16438/j.0513-4870.2022-1453
文章导航
收藏切换

利用深度学习方法设计具有全新结构的药物分子可以突破传统计算机辅助药物设计的技术瓶颈, 已经成为药物设计新技术研究的前沿, 在药物研发实践中展现了巨大潜力。本文从深度学习驱动的全新分子设计的基本原理出发, 简要介绍了深度分子生成技术和计算工具, 分析了若干代表性成功案例, 最后对该技术的未来发展方向和应用前景进行了展望。本综述将为新技术研究提供思路, 为应用该技术开展新药研发实践提供借鉴。

新药研发  /  分子信息学  /  全新药物设计  /  人工智能  /  深度学习

Design of structurally-novel drug molecules with deep learning can overcome the technical bottleneck of classical computer-aided drug design. It has become the frontier of new technique research on drug design, and has shown great potential in drug research and development practice. This review starts from the basic principles of deep learning-driven de novo drug design, goes on with the brief introduction to deep molecular generation techniques as well as computational tools and the analysis on representative successful cases, and eventually provides our perspective for future direction and application prospect about this technique. This review will provide ideas on new technique research and references for new drug research and development practice to which this technique is applied.

new drug research and development  /  molecular informatics  /  de novo drug design  /  artificial intelligence  /  deep learning
沈涛, 王冬梅, 吴松, 蒋建东, 夏杰. 基于深度学习的全新药物分子设计: 原理、工具与实践. 药学学报, 2023 , 58 (9) : 2610 -2622 . DOI: 10.16438/j.0513-4870.2022-1453
Tao SHEN, Dong-mei WANG, Song WU, Jian-dong JIANG, Jie XIA. Deep learning-based de novo drug design: principles, tools and practice[J]. Acta Pharmaceutica Sinica, 2023 , 58 (9) : 2610 -2622 . DOI: 10.16438/j.0513-4870.2022-1453
新药研发是一个周期漫长、耗资巨大的过程, 业界用“双十定律”描述这一过程, 即研发一个新药需要耗费约10年时间、约10亿美元。自Cowin Hansch于1964年提出定量构效关系(quantitative structure-activity relationship, QSAR)以来[1], 计算机辅助药物设计(computer-aided drug discovery, CADD) 作为新药研发的核心技术之一, 大大降低了新药研发成本, 加快了新药创制进程。据统计, 截至2014年, CADD技术已经辅助研发了34种上市药物[2]。经过40年的发展, 我国的CADD技术整体上也处于国际先进水平, 研发了较多上市的或者进入临床阶段的药物[3]。尽管如此, 近年来新药研发费用仍在逐年提高, 而新药上市成功率逐年降低。IQVIA的一项调研报告指出, 国外15家BigPharma企业2021年的研发费用仍高达1 330亿美元, 而上市成功率仅5%[4]。因此, 传统CADD技术似乎无法满足当今创新药物研发的高要求、新标准。若要扭转新药研发面临的困难局面, 必须进一步降低成本、缩短周期、提高成功率。为此, 传统药物设计技术及软件产品亟待升级换代。
人工智能(artificial intelligence, AI) 被描述为“制造智能机器的科学与工程”[5]。以深度学习(deep learning) 为代表的机器学习(machine learning) 方法引领了自2012年伊始的AI革命[6]。此轮热潮以计算机视觉为起点, 随后逐渐延伸到传统制造领域; 而其背后的人工神经网络(artificial neural network) 复兴主要归功于21世纪初以来的三大进展, 即持续发展的机器学习方法论、指数级增长的计算机算力、广泛可及的大数据[7]。在生物医药领域, 近年来基因组学、蛋白质组学、高通量筛选技术等的发展使得生物医药数据库得以建立和利用[8], 将数据与前沿的AI算法(深度学习) 相结合, 发展药物设计新技术, 有望突破传统CADD的技术瓶颈, 成为发展新一代药物设计技术的新动力[9]
发现具有全新结构、良好生物活性及成药性的先导化合物是创新药物研发的基础[10]。传统CADD技术如基于实体化合物库的虚拟筛选(virtual screening) 因化学空间有限、基于组合化学思维的全新药物分子设计(de novo drug design) 因合成可行性差而难以充分满足上述需求。基于深度学习的药物分子设计也分为虚拟筛选和全新分子设计两种类型。就虚拟筛选而言, 尽管深度学习可自动提取分子特征, 但其较传统CADD时代的机器学习算法(如支持向量机、随机森林) 的技术优势尚不明显: 基于深度学习的虚拟筛选的核心思想与上世纪70年代基于感知机(perceptron) 方法的QSAR建模[11]基本一致, 均属于监督学习任务; 在化合物生物活性或成药性预测任务中, 模型性能与需预先人为定义分子特征的支持向量机、随机森林等传统AI算法不相上下[12-14]。对于全新药物设计, 生成式深度学习(generative deep learning) 作为无监督机器学习的全新分支则有望突破组合化学思维、实现分子设计的多目标优化, 引领全新药物分子设计领域的重大变革[15]。2019年, Zhavoronkov等[16]利用深度生成模型在46天内完成了从靶标选择到纳摩尔级活性的高质量先导物的发现, 是生成式深度学习应用于全新分子设计领域的里程碑事件。由此, 以深度分子生成技术为代表的、基于深度学习的全新药物设计成为当前小分子药物设计技术研究与实践应用的最前沿领域。深度学习本质上是一类学习输入到输出映射的表示学习模型。与其他依赖精细特征工程的传统机器学习方法相比, 深度学习利用多层次的网络架构自动地从低维原始数据逐步抽提任务导向的高维数据特征, 生成式深度学习则基于自监督模式学习输入特征用以重建数据[17]。深度分子生成技术, 是利用生成式深度学习模型挖掘输入的化学小分子的内在特征, 充分学习分子的形成特点, 在此基础上生成骨架新颖、合成可及性良好和具有特定生物活性的虚拟分子。
自2016年第一个用于全新药物设计的深度分子生成模型问世后, 截至目前已有上百种相关计算模型发表, 也已有文献总结了相关研究工作[18, 19]。但是, 由于该技术的药物研发实践滞后于理论研究, 前人的综述并未全面概述该技术在新药研发实践中的进展。近两年来, 国内外越来越多的课题组与制药企业报道了使用基于深度学习的全新药物设计技术发现高质量先导物的应用案例, 成功实现了计算模型开发与新药研发实践的“闭环”[20]。本文首先简要概述基于深度学习的全新药物设计的基本原理, 并依据不同的生成式深度学习模型分类介绍代表性的深度分子生成工具; 以此为基础回顾分析目前已报道的基于深度学习的全新药物设计的成功案例, 以期发现深度学习应用于新药研发的基本规律, 为相关研究者提供参考。
深度分子生成模型作为基于深度学习的全新分子设计的核心, 接受低维的分子表示(molecular representation)[21]。分子信息学(molecular informatics) 中应用最广泛的分子表示方式为简化分子线性输入系统(simplified molecular input line entry system, SMILES)[22, 23]。SMILES基于分子图理论, 采用简洁而自然的语法规则将分子结构定义为字符串序列, 被广泛应用于存储化合物信息及计算化学建模, 是标准“化学语言”。近年来, 自然语言处理(natural language processing) 作为深度学习的关键应用领域在文本生成和机器翻译上发展迅猛, 而SMILES类文本的一维序列特性使其无需复杂处理便可直接用于自然语言处理建模, 其中基于一维分子表示的化学语言模型(chemical language model) 为当下最流行的深度分子生成模型[24]。尽管如此, SMILES在机器学习领域的大规模应用同样暴露出其设计理念的诸多缺陷。首先, 单个分子的SMILES表示不是唯一的, 即不同的SMILES字符串可转化为同一个分子, 这是生成分子多样性下降的重要原因; 其次, SMILES的语法规则无法保证一串由随机字符构成的SMILES序列一定能被转变成合理的分子图, 这导致了生成分子的合理度下降[25]。为解决SMILES存在的上述缺陷, Krenn等[26]针对深度学习建模特性开发了另一种一维分子表示, 被称为自参考嵌入字符串(self-referencing embedded strings, SELFIES)。SELFIES在SMILES的基础上替换了环和分支的表示方式, 并且建立了一套衍生规则显式地控制不同杂化原子的可成键数, 从而确保该字符串序列能被转为合理的分子图。尽管如此, SELFIES是否比SMILES能更有效地捕捉类药分子的内在特征仍待进一步验证。
除了一维的分子表示, 基于二维或三维的分子图(molecular graph, MG) 的表示方式因近期图神经网络(graph neural network) 的迅速发展而受到关注。以二维图为例, 图的节点被视为原子而边则被视为化学键, 不同的原子类型及价键类型则存储于张量(tensor) 之中[27]。从原理上看, 图的拓扑结构能自然地表示化学分子结构, 能更直观地反映原子间的连接信息, 因此被认为是更高效的分子表示。然而, 相较于一维文本形式的分子表示, 基于图的分子表示需占据更多的存储资源, 并且基于图数据结构的深度学习模型往往过于庞大, 因而导致计算代价昂贵; 根据已报道的基准测试, 分子图的生成式深度学习模型并未显著超越基于SMILES的模型[28]。因此, 分子图和SMILES等的分子表示都是当前深度分子生成模型的重要数据表示方式。
分子表示经处理后存储于张量之中供生成式深度学习模型学习。例如, SMILES序列在分词(tokenization) 后经独热编码(one-hot encoding) 或词嵌入(word embedding) 被转化为更高效的多维数组输入深度神经网络[29]。深度分子生成模型是AI全新药物设计算法研究中发展最迅速的领域, 这得益于近年来AI学界对发展更高效的生成式深度学习基本架构(architecture) 的巨大热情。截至目前, 已报道的深度分子生成架构包括变分自编码器(variational autoencoder, VAE)[30]、生成对抗网络(generative adversarial network, GAN)[31]、循环神经网络(recurrent neural network, RNN)[32]、强化学习(reinforcement learning, RL)[33]以及新兴的自注意力模型(Transformer)[34]和扩散模型(diffusion model)[35] (图 1)。表 1[16, 30-32, 34, 36-44]列举了各深度分子生成架构的代表性开源计算工具, 用户可以免费使用。
VAE是最早应用于分子生成的深度学习模型。VAE是自编码器(autoencoder) 的一种变体, 本质上是隐变量模型, 其使用变分推断(variational inference) 获取对数似然的下限即证据下界(evidence lower bound) 并在训练过程中使用梯度上升最大化证据下界[45]。就分子生成而言, 首先分子表示被编码器映射为低维的连续数据表示(隐变量), 随后依据预期性质使用优化算法控制该隐变量采样, 最终使用解码器重建分子表示以生成具有理想性质的分子。VAE相较于自编码器而言, 其隐空间更为平滑, 能够学习到相对可靠的连续分子表示, 因而在基于深度学习的全新药物设计算法发展的初期受到重视[46]
Gomez-Bombarelli等[30]于2016年报道了国际上第一个用于分子生成的生成式深度学习模型ChemVAE, 其编码器采用卷积神经网络(convolutional neural network) 而解码器为门控循环单元(gated recurrent unit, GRU), 并使用基于多层感知机(multilayer perceptron) 的性质预测模型对分子隐变量表示进行控制。ChemVAE初步展现了生成式深度学习模型应用于全新分子设计的潜力, 然而ChemVAE需预先清除不合理SMILES以显式地控制生成分子的合理度, 表明模型学习隐空间分子表示的能力欠佳。后续关于VAE的改进集中于分子表示及隐变量采样优化方面, 而Ragoza等[37]的最新工作将VAE的应用范围拓展至三维分子生成领域以进行基于结构的全新分子设计。该架构的编码器分为输入端与条件端, 前者接受三维格点化的受体-配体复合物作为输入, 后者则将三维格点化的单独受体作为条件限制输入。解码器从编码器两端拼接的隐变量分布中采样以输出配体的三维格点。最后通过原子拟合算法及键推理算法从三维格点中构建合理分子。作者使用突变的莽草酸激酶(shikimate kinase) 测试该模型的条件生成能力。结果显示, 多个残基突变均能引起生成的分子性质的显著变化, 表明在三维分子的生成过程中模型成功利用了受体-配体间相互作用信息。
GAN由Goodfellow等[47]于2014年提出, 其变体DCGAN、WGAN、SeqGAN在图像及文本生成方面的优秀表现迅速取代了VAE模型[48]。GAN基于博弈学习理论包含两个子网络: 生成网络(generator, G) 和判别网络(discriminator, D)。G负责生成接近真样本的伪样本, D则通过观察真样本与伪样本的区别试图将二者区分, 二者互相博弈最终达到平衡, 即纳什均衡(Nash equilibrium)。
目标强化的生成对抗网络(objective reinforced generative adversarial network, ORGAN) 是GAN应用于AI全新分子设计的代表性算法[31]。ORGAN的核心在于通过基于REINFORCE算法[49]的强化学习架构以优化生成的分子的油水分配系数(oil/water partition coefficient, logP)、类药定量评估(quantitative estimate of drug-likeness, QED) 分数[50]及合成可及性得分(synthetic accessibility score, SAscore)[51]。ORGAN相较于VAE模型有更高的生成分子合理度, 与含有优化策略的RNN模型相比则有更高的性质优化成功率。
RNN是目前最常见的用于序列建模的深度学习架构。相较于传统人工神经网络, 其引入了隐状态(hidden state) 以保存先前时间序列至当前时间步的信息, 通过循环计算在下一个时间步进行输出, 因此RNN具有序列敏感的“记忆功能”[52]。RNN作为子网络也常在VAE及GAN等架构中出现, 此处仅讨论使用RNN进行语言建模的深度分子生成模型[24]。化学语言模型将特征向量化的一维分子表示(即序列) 作为输入, 通过训练模型实现下一个预测分词(token) 的似然(likelihood) 的最大化; 训练好的模型可输出完整的字符串, 经标准化处理后转化为合理的化学分子。化学语言模型采用深层RNN以获得更高维的类药分子特征向量, 并使用隐状态可控的基本RNN单元变体长短期记忆网络(long short-term memory, LSTM) 或GRU以解决网络预测长程依赖序列时出现的梯度消失(vanishing gradient problem) 或梯度爆炸问题(exploding gradient problem)[53]。化学语言模型在经过基于大规模化学小分子数据的预训练后, 便可随机生成合理的分子。而生成满足特定理化性质或结构特征的类药分子, 需进一步采用目标优化策略, 其中最常用的两种策略为迁移学习(transfer learning) 和强化学习。迁移学习在化学语言模型中的应用指使用更为聚焦的小型分子库(focused library) 微调已掌握类药分子基本特征的预训练模型, 从而使模型学习到小型分子库的概率分布; 若该小型分子库包含某一特定药理活性的分子, 则经过迁移学习微调的化学语言模型可以生成具有潜在药理活性的全新分子结构[38]。相较于迁移学习这种数据驱动的策略, 整合了强化学习模型的化学语言模型的核心则是依据奖励目标建立一个奖励函数对生成的分子进行实时打分, 基于策略的强化学习模型常使用策略梯度算法(policy gradient algorithm) 实现全局累积奖励的最大化, 即最大程度地满足用户所预设的分子特征[54]
Gupta等[38]和Segler等[32]率先于2018年同时报道了基于RNN的深度分子生成工具并提出了化学语言模型的概念。他们在模型构建中均使用了多层LSTM、丢弃(dropout) 法及迁移学习策略以增强网络性能和拓宽应用范围。Gupta等[38]基于迁移学习策略成功构建了针对过氧化物酶体增殖物激活受体γ (peroxisome proliferator-activated receptor gamma, PPARγ) 与胰蛋白酶的虚拟分子库, 并且验证了化学语言模型同样可应用于先导结构优化。一方面将完成预训练的化学语言模型在目标分子上进行微调, 使得模型可输出与之结构有一定相似度但部分骨架跃迁的全新分子; 另一方面将先导结构的核心骨架作为起始序列输入模型, 模型可在此基础上进行结构修饰, 实现片段生长。Segler等[32]借鉴药物化学中“设计-合成-测试”的循环优化流程, 设计了基于化学语言模型的AI全新分子设计循环: 首先, 使用预训练的RNN模型随机采样一批合理分子; 之后使用基于机器学习方法的QSAR模型对采样的分子进行分类, 获得标注为活性的生成分子; 最后将预测为活性的分子重新输入网络进行微调。通过上述过程的循环, 模型可不依赖于具体数据而学习到具有特定活性的分子的特征, 从而解决了反向QSAR (inverse QSAR) 问题。以抗金葡菌(Staphylococcus aureus) 的化合物为例, RNN模型仅经过8次迭代便可召回QSAR模型测试集中6%的分子, 并且共获得6万多个预测为活性的全新分子, 展现了RNN生成潜在活性分子的强大能力。
强化学习通过智能体(agent) 与环境(environment) 交互, 学习采取行动(action) 的策略以获取累积最大奖励(reward), 其通常与深度学习框架结合以构建深度强化学习(deep reinforcement learning) 模型, 目前已在围棋等动态决策任务上展现出远超人类的性能[55]。前文所述的部分深度分子生成模型结合了强化学习方法以实现条件分子生成, 在此不再赘述。事实上, 强化学习作为主体AI框架可直接应用于深度分子生成。例如, Zhou等[40]开发的MolDQN模型基于深度Q学习(deep Q-learning) 算法, 将候选分子的修饰优化视为马尔科夫决策过程(Markov decision process), 通过显式地融入已知的化学知识控制分子图的构建, 因而此模型无需任何预训练便可从头生成分子。
DeepLigBuilder是强化学习在基于结构的三维分子生成技术中的最新进展[56]。该模型包括L-Net和强化学习两部分, 前者由编码三维分子图的消息传递网络和进行图编辑决策的掩蔽自编码器组成, 用于非条件性地生成类药分子三维构象。后者则基于蒙特卡洛树搜索(Monte Carlo tree search) 这一强化学习策略函数, 以对接得分作为奖励实现受体结合位点生成三维配体的在线优化。作者模拟现实应用场景针对SARS-CoV-2主蛋白酶Mpro进行了三维全新分子设计。结果显示, DeepLigBuilder不仅能够基于已知片段生成具有更高对接亲和力的分子, 而且能够充分利用受体结合位点特征从头生成高质量的虚拟三维配体。
基于Transformer的大规模预训练语言模型是当前人工神经网络复兴以来自然语言处理领域的最大进展, 其在文本分类、文本生成与机器翻译等多个任务上取得了最优异效果, 并被迅速应用于AI全新分子设计之中[57]。最早的Transformer模型为编码器-解码器架构, 由基于注意力机制(attention mechanism) 的模块堆叠而成[58], 用以取代传统的基于RNN的序列到序列(sequence-to-sequence, Seq2Seq) 的模型。Transformer模型的最新进展分别为基于编码器砌块构建的来自Transformers的双向编码器表示(bidirectional encoder representation from Transformers, BERT) 模型[59]和基于解码器砌块构建的生成预训练Transformer (generative pre-trained Transformer, GPT) 模型[60], 二者都是上下文敏感的预训练模型, 通过微调即可应用于多种下游任务。Transformer模型使用的多头自注意力(multi-head self-attention) 机制使其能够并行计算并且学习到特定分词在完整序列中的长程依赖, 因此基于Transformer的深度分子生成模型在生成分子的合理度上有较大提升。另外, 其在长序列学习中的优势也使得对基因序列和蛋白质一级结构序列的建模成为可能, 从而能够在生成过程中更多地纳入靶标特征[61]
Wang等[41]发展的多条件限制分子生成(multi-constraint molecular generation, MCMG) 模型系统性地探索了Transformer在条件深度分子生成中的适用性。MCMG的核心在于构建了基于GPT架构的条件Transformer: 带有条件限制的张量与SMILES分词嵌入张量拼接共同输入网络进行训练, 其原理与化学语言模型类似; 之后模型通过知识蒸馏(knowledge distillation) 策略将条件Transformer已学习到的化学语言模型条件生成规则迁移至RNN模型以降低模型复杂度, 从而可进一步使用强化学习以诱导分子生成方向。MCMG在MOSES基准测试中的多个指标上超越了RNN模型, 并且在任务导向的分子生成评价测试中以89.26%的成功率领先目前基于RNN和强化学习的最佳模型REINVENT平台[62], 展现了Transformer模型在深度分子生成的巨大潜力。
Zheng等[42]将Transformer模型运用于分子骨架跃迁并致力于纳入更多生物学信息。该研究小组开发的DeepHop工具借鉴了分子-分子翻译的Seq2Seq框架, 通过输入二维不相似而三维相似的分子对进行模型训练。编码器同时接受分子一维的SMILES序列嵌入和三维图卷积的分子构象嵌入作为输入, 并且使用了预训练的蛋白序列嵌入以增强靶标信息。在Pim-1激酶抑制剂的回顾性测试例中, 对于3纳摩尔级的先导物, DeepHop生成的骨架跃迁分子新颖率均超过50%, 并且在最佳案例中高达69.7%的生成分子拥有更佳的对接亲和力, 表明了该模型在药物分子设计实际应用场景中的巨大优势。
Yang等[63]于2020年提出的去噪扩散模型(denoising diffusion probabilistic model, DDPM) 引领了当下生成式深度学习的热潮, 其核心算法在于使用两条马尔科夫链分别对数据进行前向噪声扩散和反向去噪重建。连续扩散模型(continuous diffusion model) 包含了近期此类模型开发的重要进展, 如级联扩散模型[64]和扩散语言模型[65], 前者用于生成高分辨率的图像, 后者则将扩散模型拓展至文本生成领域。虽然该类框架的计算理论研究尚未完善, 但是目前已有工作报道其在三维分子生成领域中的应用。例如, Hoogeboom等[43]利用E(3) 等变图神经网络稳定了DDPM的噪声扩散过程, 三维分子生成的稳定性与独特性超越了VAE模型。Schneuing等[44]在该工作基础上开发了DiffSBDD, 其将蛋白质口袋的等变图表示作为条件以控制三维分子生成。基准测试显示该模型可针对给定口袋产生更高对接亲和力的配体。相较于前述深度学习框架, 连续扩散模型在深度分子生成技术中的应用仍属起步阶段, 其模型性能、功能广度和湿实验应用可行性将在未来进一步拓展。
本节将基于生物靶标类别解析基于深度分子生成技术发现高质量药物先导结构的代表性成功案例(表 2), 重点关注AI及CADD方法运用的宏观策略以期发现案例间的共性。
核受体(nuclear receptors, NRs) 是一类受激素调控的转录因子, 由于核受体具有大量的化学生物学数据, 针对核受体的AI全新分子设计成为第一个成功案例[66, 67]。Merk等[68]基于Gupta等开发的化学语言模型, 使用迁移学习策略生成维甲酸类X受体(retinoid X receptors, RXRs) 与过氧化物酶体增殖物激活受体(peroxisome proliferators-activated receptors, PPARs) 激动剂的虚拟分子库。该计算模型的构建有两大特点: ①此类靶标激动剂均为脂肪酸类似物, 由此团队以“-COOH”为起始片段进行分子生成; ②针对性地使用了基于分子三维形状与电荷状态描述符的相似性搜索与QSAR模型SPiDER进行虚拟筛选[69]。作者采样了1 000个SMILES作为起始分子, 最后聚焦为49个多样性分子并选取了其中5个分子进行湿实验验证。通过使用商业可及的原料和4步反应合成了上述化合物, 通过杂交报告基因法确定了2个化合物对RXRs与PPARγ均有双重激动活性, 其中对激动RXRγ的EC50值最高达到0.06 μmol·L-1。此项工作首次证明了化学语言模型在实际应用时的高度灵活性与可行性, 引领了此类模型的深入发展。同时该工作也具有开创意义, 其确立了生成式深度学习计算模型与实验相结合的基本思路, 如使用迁移学习方法构建特定靶标的虚拟分子库及使用定制的虚拟筛选策略进一步缩小目标范围等。
多巴胺2型受体(dopamine D2 receptor, D2R)、5-羟色胺1A受体(5-HT1A receptor, 5-HT1AR) 和5-羟色胺2A受体(5-HT2A receptor, 5-HT2AR) 是治疗精神分裂症的重要靶标[70]。作用于D2R单靶点的第一代抗精神病药物锥体外系不良反应明显, 而同时作用于5-HT2AR与D2R等受体的第二代药物因不良反应小且具有认知改善功能而受到临床青睐。Tan等[71]率先探索了深度分子生成技术在多靶点抗精神病药物设计和发现中的应用。该研究同样使用化学语言模型作为基本架构, 其中用于迁移学习的初始数据集为EC50或IC50低于1 μmol·L-1的多靶点活性分子, 包括5-HT1AR激动剂和D1R/D2R/5-HT2AR拮抗剂。此部分活性数据同时被用于构建一个基于多任务深度神经网络(multitask deep neural network) 的QSAR模型对生成分子进行活性预测。在化学语言模型训练的过程中, 每个周期结束后会生成一批分子, 其中被QSAR模型预测为高活性的类药分子被用于扩展迁移学习数据集以进行下个周期的迭代。最终该模型可生成具有多靶点活性的抗精神病候选化合物。在该案例中, 作者共采样10 000个分子作为虚拟分子库, 通过将中枢神经系统药物的重要理化特征如logP、血脑分配系数和拓扑极性表面积(topological polar surface area) 作为筛选条件并结合上述QSAR模型的预测结果, 确定了5个虚拟候选分子, 最终选择了化合物3进行实验验证。化合物3对D2R与5-HT2AR的IC50分别为216和1.64 nmol·L-1, 展现出了非典型药物的特点; 同时对5-HT1AR产生EC50值为0.51 nmol·L-1的高度激活效应, 与目前已报道的药物有所不同。该团队进一步设计和合成了化合物3的10个衍生物, 并对其中6个进行了实验验证。在苯环己哌啶诱导的运动增多模型[phencyclidine(PCP)-induced hyperactivity model] 中, 仅在0.1~0.3 mg·kg-1剂量下, 化合物8即可逆转小鼠症状而不影响其自主活动, 同剂量的不良反应明显少于阿立哌唑。该项工作通过构建疾病导向的迁移学习数据集及使用反向QSAR方法拓展了化学语言模型的应用范围, 展现了AI全新分子设计在多靶点药物开发中的巨大潜力。
蛋白激酶是靶向药物设计与发现中最重要的生物靶标之一。截至2022年, 美国FDA已批准了71个小分子激酶抑制剂用于治疗恶性肿瘤、组织纤维化和炎症等疾病[72]。蛋白激酶的生物学功能研究全面[73], 相关的小分子生物活性数据在诸多化学信息数据库中均有详细记录, 因此相关先导物的设计与发现成为了AI全新分子设计中最热门的应用场景。
2018年, Putin等[74]率先报道了使用基于GAN的生成式深度学习模型发现全新激酶抑制剂。该团队借鉴ORGAN系列工作提出了对抗阈值神经计算机(adversarial threshold neural computer, ATNC)。此模型通过引入微分神经计算机模块与对抗阈值模块以提高神经网络对长序列SMILES的建模能力。在实验验证中, 来自Thomson Integrity数据库的约3万个已知活性的激酶抑制剂作为训练集被输入网络, 训练完毕后的ATNC输出了约3万个候选分子供后续筛选。首先, 作者使用了常见的药物化学性质过滤符以保留类药分子。为确保生成分子的结构新颖性并可突破现有专利保护, 剔除了与训练集分子指纹相似性超过0.7的生成分子。最后, 通过SciFinder与ChEMBL数据库检索确保了剩余分子尚无激酶活性报道。该团队购买了与50个候选虚拟分子结构最为接近的商业可及分子进行激酶抑制活性测试, 其中7个化合物在10 μmol·L-1时对不同的激酶均有超过50%的抑制率, 并且2个化合物表现出较好的选择性。该工作在生成策略设计与化学实验结合上缺陷较为明显, 如未针对特定靶标进行训练且未考虑合成全新化合物进行评价等, 但其仍是早期深度分子生成模型在先导物发现中的成功应用, 揭示了深度分子生成在药物发现领域的巨大潜力。
正如前文所提及的, 一年后Zhavoronkov等[16]简要介绍了基于VAE架构的生成式深度学习模型在受体酪氨酸激酶-盘状结构域受体1 (discoidin domain receptor 1, DDR1) 抑制剂设计和发现中的应用, 产生了较大的社会影响。该团队从构建生成张量化强化学习(generative tensorial reinforcement learning, GENTRL) 模型到最终获得设计的6个化合物仅耗时21天, 其中化合物12均满足Lipinski原则, 对DDR1的IC50分别为10和21 nmol·L-1。尽管此项工作被认为是全新分子生成领域的极为重要的进展, 然而有两处明显不足: 一是有批评指出该化合物与已上市的普纳替尼(ponatinib) 在化学结构上高度相似, 并且在激酶谱测试完整性上存在缺陷[75]; 二是该文章并没有报道该结构的体内药效数据。不同于通过生成式深度学习模型直接生成全新分子, Tan等[76]以课题组前期发现的成纤维细胞生长因子受体(fibroblast growth factor receptors, FGFR) 抑制剂DC-1的关键母核为基本骨架, 进行了针对DDR1的AI全新分子改造。该团队开发了一种基于注意力机制的Seq2Seq模型用以对母核进行骨架修饰, 首先用匹配分子对算法将ChEMBL数据库来源的902个DDR/FGFR小分子抑制剂分割重组为36亿个骨架片段-装饰片段分子对, 然后输入网络进行训练; 训练完毕的模型即可在母核的指定位点进行衍生。生成式深度学习模型对吡唑并[3, 4-d]哒嗪酮骨架进行双位点修饰, 共产生19 929个分子。该团队使用了自研的KinomeX激酶谱预测平台进行选择性预测, 并利用Glide分子对接作进一步虚拟筛选, 最终选取了两个化合物进行合成及生物活性评价。结果显示, 化合物2是一种强的DDR1抑制剂(IC50 = 10.6 nmol·L-1); 激酶谱测试显示, 在100 nmol·L-1浓度下的化合物2仅对原肌球蛋白受体激酶C (tropomyosin receptor kinase C) 产生55%的抑制率, 表现出高选择性。该团队还应用葡聚糖硫酸钠诱导的小鼠溃疡性结肠炎[dextran sulfate sodium (DSS)-induced inflammatory bowel disease] 模型测试了该化合物的体内抗炎活性, 结果显示其与阳性药DDR1-IN-1活性相当。该项工作全面探索了如何综合应用AI全新分子设计技术与传统CADD策略进行激酶全新抑制剂发现, 并且弥补了Zhavoronkov等[16]工作的不足。
近期Li等[77]通过总结以上成功经验展现了基于深度分子生成模型设计和发现蛋白激酶抑制剂的范式。在该研究工作中, 生成式深度学习模型从训练与采样两个角度出发, 基于隐状态正则化与线性插值策略改进了目前最为成功的化学语言模型。该模型以生成受体相互作用蛋白激酶1 (receptor-interacting protein kinase 1, RIPK1) 的虚拟分子库为目的, 首先以ZINC数据库中的约1 600万个有机小分子作为输入对模型进行预训练, 之后以文献与专利报道的1 030个已知RIPK1抑制剂作为数据集进行迁移学习, 最终采样了79 323个候选分子。随后他们使用传统CADD策略进一步缩小化学空间: ①基于拓扑结构与物理化学性质的类药性过滤; ②基于药效团模型和分子对接方法的虚拟筛选。最终该团队从得分最高的前50个化合物中选取了8个全新分子作为苗头化合物进行化学合成及生物活性评价。此篇文献集中报道了对化合物RI-962的研究, 该化合物对RIPK1具有选择性抑制, 活性达到纳摩尔级(IC50 = 5.9 nmol·L-1), 对于体外细胞坏死的保护作用与II期临床药物GSK3145095相当, 而体内抗炎药效则优于GSK3145095。作者团队解析的RIPK1-RI-962晶体复合物结构显示该先导物为Ⅱ型激酶抑制剂, 其末端苯环结构伸入RIPK1特异的变构位点, 产生大量疏水相互作用, 充分揭示了高活性与高选择性。此项工作的先导物设计和发现流程具有高度的可拓展性, 其临床前研究深入而全面, 是目前基于深度学习的全新药物设计的典型应用案例。
在当前新药研发面临巨大挑战的大背景下, 药物设计技术亟需升级换代。而近年来兴起的AI革命恰逢其时, 为新一代药物设计技术发展注入了新的动力。除了高通量虚拟筛选为主要应用场景外[81], 生成式深度学习模型因有望突破传统全新分子设计的技术瓶颈, 成为当前药物设计的最前沿领域。
小分子药物的发现过程通常被描述为对类药化学空间的探索[82]。传统虚拟筛选与全新药物分子设计作为CADD领域中的重要概念代表了两种探索模式。前者预先设定已知的类药空间而后利用计算方法评估空间内候选分子的生物活性和类药性, 后者则是基于优化算法不断变化分子结构以满足预期活性或性质, 本质上是对未知空间的探索。实际上, 当前的基于深度学习的全新药物分子设计策略是对以上两种模式的整合与发展。一方面, 目前主流的化学语言模型的应用范式为基于迁移学习策略生成特定靶点或活性的虚拟分子库, 随后使用传统虚拟筛选方法进一步聚焦化学空间以选取高质量的潜在活性分子进行实验验证; 另一方面, 用于骨架跃迁或片段连接的生成式深度学习模型则可实现对先导结构的优化。事实上, 传统CADD所采用的全新分子设计同样是基于以上两种模式[83], 然而其产生的虚拟分子的验证可行性远低于目前的深度生成方法, 这是由其“特征驱动”的本质所决定的: 基于特征分子砌块与特定反应规则构建的分子既无法彻底地探索未知化学空间而获得足够新颖的结构, 也无法保证虚拟分子接近已知类药分子的真实特征[84]。与之相比, 深度分子生成方法的最大优势在于其完全由“数据驱动”。以字符级的化学语言模型为例, 模型首先从最基本的化学数据结构SMILES中自动抽提特征, 通过学习到每个分词在整个序列中的概率分布, 在原子级的维度上构建合理分子, 再使用迁移学习或强化学习策略微调模型生成分子的方向, 使得生成的分子具有满足特定活性或性质的结构特征。尽管目前用于生成重定向的小分子库大多针对某一特定靶标或蛋白家族, 将其拓展到具有表型活性的更大分子库在模型理论上同样可行, 并且借助高性能计算设备的充分采样能够更为全面地覆盖某一类型的化学空间, 进一步提升发现全新分子的概率。
基于深度学习的全新药物设计作为一种“数据驱动”的先导结构设计和发现技术, 化学分子数据的来源与处理方法直接决定了生成式深度学习模型生成的分子的品质[85]。目前深度分子生成最常使用的数据库包括ZINC[86]、ChEMBL[87]和GDB[88], 均为开放获取的非商业化合物库[8]。就预训练而言, 模型首先需要从有机小分子上学习基本的“化学语言”以确保能生成合理的化学分子, 而学界对于预训练数据量大小、数据性质、数据质量及具体训练方式仍缺乏统一的标准。通常认为, 预训练需要输入数以百万乃至千万计的小分子进行充分训练, 此种策略不仅计算代价昂贵, 而且对数据处理策略提出了极高的要求。最新研究显示, 化学语言模型在小数据场景下结合数据扩增和温控采样技术同样可输出有效分子[89]。除此之外, 化学数据性质的选择也至关重要。ChEMBL作为最大的小分子生物活性数据库, 本身包含了大量高活性的类药分子, 相比之下ZINC则强调化合物的商业可及性, GDB中的分子则完全基于计算策略虚拟构建, 是否在预训练过程中便强调化学空间的类药性还是更加注重分子的多样性仍值得深入探讨。
用于全新药物分子设计的生成式深度模型与其他计算模型一样需要进行基准测试以评估其性能。MOSES[28]与GuacaMol[90]是两种主流的基准测试平台, 前者强调对模型生成的分子的合理度、新颖性及骨架多样性等通用类药指标进行测试, 后者则通过定义一系列任务以评价模型的多目标优化能力。尽管上述基准测试平台已被广泛应用于评估常用架构的性能, 然而其过度关注计算性能指标而忽略了实践中深度生成模型时的复杂需求, 如生成分子的逆合成可行性等。药物分子设计与发现的复杂性决定了标准化评测方法仍需进一步发展。
本文所涉及的成功案例简要展现了基于深度学习的全新药物分子设计技术近年来在实践中从尝试探索逐步走向规范化的历程, 然而相较于计算模型的井喷式发展, 其仍处于初级发展阶段, 即该技术尚未实现模型开发与生产实践的“闭环”。在目前已公开报道的11例应用实践中[16, 36, 39, 68, 71, 74, 76-80], 仅有4例报道了先导物的体内药效数据[71, 76-78], 表明该新兴技术应用于实践的具体流程仍存在巨大的提升空间。在这些成功案例中, 可以看到传统CADD技术仍然发挥着不可替代的作用。基于QED与SAscore的初步过滤确保了所生成的分子具备基本的类药性; 基于分子指纹的相似性计算辅助剔除了低知识产权的结构冗余分子; 基于QSAR、药效团模型和分子对接的虚拟筛选方法对于缩小化学空间极为重要, 可辅助选择高质量分子。然而, 传统CADD技术仍有关键科学问题需要解决。一方面, 关于蛋白受体结合位点具体残基的柔性与结晶水的取舍仍需特别考虑, 蛋白-配体亲和力打分函数的精度与结合自由能计算的速度也远未达预期[91]; 另一方面, 基于机器学习的QSAR模型及分子对接模型面临着可用数据量小、数据内部存在偏好性的问题, 其性能的进一步提升受到限制[92]。在CADD技术进步的基础上, 将传统CADD策略与多任务深度神经网络等AI模型相融合有望进一步提升先导物的发现效率, 体现CADD与AI全新药物分子设计的交互与融合。
基于深度学习的全新药物设计是近10年来AI在传统学科领域最重大的突破之一[93]。一方面得益于计算机和信息领域对生成式深度学习的高度关注, 更高效的数据表示策略与更先进的生成式深度学习架构被迅速应用于深度分子生成领域, 相关模型开发与基准测试的规范正在逐步确立; 另一方面传统药物化学领域也在积极尝试应用这一新技术实现降本增效, 相关基本应用流程已经初步确立。新药创制是一门多学科交叉、受新兴技术推动的复杂工程, AI在该领域的应用尚属起步阶段。期待深度学习驱动的全新药物设计进一步助力药物发现, 在不远的将来成为生物医药的新范式。
作者贡献: 蒋建东提出论文主题、对论文撰写给予指导; 沈涛和夏杰进行文献检索和整理、确定论文框架和负责论文撰写; 王冬梅和吴松修改论文。
利益冲突: 所有作者声明不存在利益冲突。
  • 中国医学科学院医学与健康科技创新工程(2021-I2M-1-069)
  • 国家重大新药创制科技重大专项(2018ZX09711001-012-003)
参考文献 引证文献
排序方式:
[1]
Cherkasov A, Muratov EN, Fourches D, et al. QSAR modeling: where have you been? where are you going to?[J]. J Med Chem, 2014, 57: 4977-5010.
[2]
Ghosh AK, Gemma S. Structure-based Design of Drugs and Other Bioactive Molecules: Tools and Strategies[M]. New Jersey: John Wiley & Sons, 2014: 12-13.
[3]
Tan XQ, Xiong JC, Zhu TF, et al. Development of drug design in China: 40 years of achievements[J]. Sci Sin Vit (中国科学: 生命科学), 2019, 49: 1375-1394
[4]
Aitken M, Kleinrock M. Global Trends in R & D: Overview through 2021 [EB/OL]. (2022-02-10) [2022-12-24]. https://www.iqvia.com/insights/the-iqvia-institute/reports/global-trends-in-r-and-d-2022.
[5]
Hamet P, Tremblay J. Artificial intelligence in medicine[J]. Metabolism, 2017, 69: S36-S40.
[6]
Krizhevsky A, Sutskever I, Hinton GE. Imagenet classification with deep convolutional neural networks[J]. Commun ACM, 2017, 60: 84-90.
[7]
LeCun Y, Bengio Y, Hinton G. Deep learning[J]. Nature, 2015, 521: 436-444.
[8]
Humbeck L, Koch O. What can we learn from bioactivity data? chemoinformatics tools and applications in chemical biology research[J]. ACS Chem Biol, 2017, 12: 23-35.
[9]
Rifaioglu AS, Atas H, Martin MJ, et al. Recent applications of deep learning and machine intelligence on in silico drug discovery: methods, tools and databases[J]. Brief Bioinform, 2018, 20: 1878-1912
[10]
Sun DX, Gao W, Hu HX, et al. Why 90% of clinical drug development fails and how to improve it?[J]. Acta Pharm Sin B, 2022, 12: 3049-3062.
[11]
Hiller SA, Golender VE, Rosenblit AB, et al. Cybernetic methods of drug design. Ⅰ. Statement of the problem—the perceptron approach[J]. Comput Biomed Res, 1973, 6: 411-421.
[12]
Yang K, Swanson K, Jin W, et al. Analyzing learned molecular representations for property prediction[J]. J Chem Inf Model, 2019, 59: 3370-3388.
[13]
van Tilborg D, Alenicheva A, Grisoni F. Exposing the limitations of molecular machine learning with activity cliffs[J]. J Chem Inf Model, 2022, 62: 5938-5951.
[14]
Janela T, Bajorath J. Simple nearest-neighbour analysis meets the accuracy of compound potency predictions using complex machine learning models[J]. Nat Mach Intell, 2022, 4: 1246-1255.
[15]
Kim J, Park S, Min D, et al. Comprehensive survey of recent drug discovery using deep learning[J]. Int J Mol Sci, 2021, 22: 9983.
[16]
Zhavoronkov A, Ivanenkov YA, Aliper A, et al. Deep learning enables rapid identification of potent DDR1 kinase inhibitors[J]. Nat Biotechnol, 2019, 37: 1038-1040.
[17]
Vanhaelen Q, Lin YC, Zhavoronkov A. The advent of generative chemistry[J]. ACS Med Chem Lett, 2020, 11: 1496-1505.
[18]
Xu YJ, Lin KJ, Wang SW, et al. Deep learning for molecular generation[J]. Future Med Chem, 2019, 11: 567-597.
[19]
Elton DC, Boukouvalas Z, Fuge MD, et al. Deep learning for molecular design—a review of the state of the art[J]. Mol Syst Des Eng, 2019, 4: 828-849.
[20]
Wang M, Wang Z, Sun H, et al. Deep learning approaches for de novo drug design: an overview[J]. Curr Opin Struct Biol, 2022, 72: 135-144.
[21]
Winter R, Montanari F, Noé F, et al. Learning continuous and data-driven molecular descriptors by translating equivalent chemical representations[J]. Chem Sci, 2019, 10: 1692-1701.
[22]
Weininger D. SMILES, a chemical language and information system. 1. Introduction to methodology and encoding rules[J]. J Chem Inf Comput Sci, 1988, 28: 31-36.
[23]
Weininger D, Weininger A, Weininger JL. SMILES. 2. Algorithm for generation of unique SMILES notation[J]. J Chem Inf Comput Sci, 1989, 29: 97-101.
[24]
Skinnider MA, Stacey RG, Wishart DS, et al. Chemical language models enable navigation in sparsely populated chemical space[J]. Nat Mach Intell, 2021, 3: 759-770.
[25]
Krenn M, Ai Q, Barthel S, et al. SELFIES and the future of molecular string representations[J]. Patterns, 2022, 3: 100588.
[26]
Krenn M, Häse F, Nigam A, et al. Self-referencing embedded strings (SELFIES): a 100% robust molecular string representation[J]. Mach Learn Sci Technol, 2020, 1: 045024.
[27]
Li Y, Zhang L, Liu Z. Multi-objective de novo drug design with conditional graph generative model[J]. J Cheminform, 2018, 10: 33.
[28]
Polykovskiy D, Zhebrak A, Sanchez-Lengeling B, et al. Molecular sets (MOSES): a benchmarking platform for molecular generation models[J]. Front Pharmacol, 2020, 11: 565644.
[29]
Pillai N, Dasgupta A, Sudsakorn S, et al. Machine learning guided early drug discovery of small molecules[J]. Drug Discov Today, 2022, 27: 2209-2215.
[30]
Gómez-Bombarelli R, Wei JN, Duvenaud D, et al. Automatic chemical design using a data-driven continuous representation of molecules[J]. ACS Cent Sci, 2018, 4: 268-276.
31
Guimaraes GL, Sanchez-Lengeling B, Outeiral C, et al. Objective-reinforced generative adversarial networks (ORGAN) for sequence generation models [J]. arXiv, 2017: 1705.10843.
[32]
Segler MHS, Kogej T, Tyrchan C, et al. Generating focused molecule libraries for drug discovery with recurrent neural networks[J]. ACS Cent Sci, 2018, 4: 120-131.
[33]
Olivecrona M, Blaschke T, Engkvist O, et al. Molecular de-novo design through deep reinforcement learning[J]. J Cheminform, 2017, 9: 48.
[34]
Yang Y, Zheng S, Su S, et al. SyntaLinker: automatic fragment linking with deep conditional transformer neural networks[J]. Chem Sci, 2020, 11: 8312-8322.
35
Wu L, Gong C, Liu X, et al. Diffusion-based molecule generation with informative prior bridges [J]. arXiv, 2022: 2209.00865.
[36]
Godinez WJ, Ma EJ, Chao AT, et al. Design of potent antimalarials with generative chemistry[J]. Nat Mach Intell, 2022, 4: 180-186.
[37]
Ragoza M, Masuda T, Koes DR. Generating 3D molecules conditional on receptor binding sites with deep generative models[J]. Chem Sci, 2022, 13: 2701-2713.
[38]
Gupta A, Müller AT, Huisman BJH, et al. Generative recurrent networks for de novo drug design[J]. Mol Inform, 2018, 37: 1700111.
[39]
Li XY, Xu YQ, Yao HQ, et al. Chemical space exploration based on recurrent neural networks: applications in discovering kinase inhibitors[J]. J Cheminform, 2020, 12: 42.
[40]
Zhou Z, Kearnes S, Li L, et al. Optimization of molecules via deep reinforcement learning[J]. Sci Rep, 2019, 9: 10752.
[41]
Wang J, Hsieh CY, Wang M, et al. Multi-constraint molecular generation based on conditional transformer, knowledge distillation and reinforcement learning[J]. Nat Mach Intell, 2021, 3: 914-922.
[42]
Zheng SJ, Lei ZR, Ai HT, et al. Deep scaffold hopping with multimodal transformer neural networks[J]. J Cheminform, 2021, 13: 87.
[43]
Hoogeboom E, Satorras VG, Vignac C, et al. Equivariant diffusion for molecule generation in 3D [C] // Proceedings of the 39th International Conference on Machine Learning. Maryland: International Machine Learning Society, 2022: 8867-8887.
44
Schneuing A, Du YQ, Harris C, et al. Structure-based drug design with equivariant diffusion models [J]. arXiv, 2022: 2210.13695.
[45]
Doersch C. Tutorial on variational autoencoders [J]. arXiv, 2016: 1606.05908.
[46]
Blaschke T, Olivecrona M, Engkvist O, et al. Application of generative autoencoder in de novo molecular design[J]. Mol Inform, 2018, 37: 1700123.
[47]
Goodfellow I, Pouget-Abadie J, Mirza M, et al. Generative adversarial networks[J]. Commun ACM, 2020, 63: 139-144.
[48]
Creswell A, White T, Dumoulin V, et al. Generative adversarial networks: an overview[J]. IEEE Signal Process Mag, 2018, 35: 53-65.
[49]
Recht B. A tour of reinforcement learning: the view from continuous control[J]. Annu Rev Control Robot Auton Syst, 2019, 2: 253-279.
[50]
Bickerton GR, Paolini GV, Besnard J, et al. Quantifying the chemical beauty of drugs[J]. Nat Chem, 2012, 4: 90-98.
[51]
Ertl P, Schuffenhauer A. Estimation of synthetic accessibility score of drug-like molecules based on molecular complexity and fragment contributions[J]. J Cheminform, 2009, 1: 8.
52
Graves A. Generating sequences with recurrent neural networks [J]. arXiv, 2013: 1308.0850.
[53]
Yu Y, Si X, Hu C, et al. A review of recurrent neural networks: LSTM cells and network architectures[J]. Neural Comput, 2019, 31: 1235-1270.
[54]
Popova M, Isayev O, Tropsha A. Deep reinforcement learning for de novo drug design[J]. Sci Adv, 2018, 4: eaap7885.
[55]
Silver D, Huang A, Maddison CJ, et al. Mastering the game of go with deep neural networks and tree search[J]. Nature, 2016, 529: 484-489.
[56]
Li Y, Pei J, Lai L. Structure-based de novo drug design using 3D deep generative models[J]. Chem Sci, 2021, 12: 13664-13675.
[57]
Yang L, Yang G, Bing Z, et al. Transformer-based generative model accelerating the development of novel BRAF inhibitors[J]. ACS Omega, 2021, 6: 33864-33873.
[58]
Vaswani A, Shazeer N, Parmar N, et al. Attention is all you need [C] // Advances in Neural Information Processing Systems. California: Curran Associates, 2017: 5998–6008.
59
Devlin J, Chang MW, Lee K, et al. Bert: pre-training of deep bidirectional transformers for language understanding [J]. arXiv, 2018: 1810.04805.
[60]
Qiu XP, Sun TX, Xu YG, et al. Pre-trained models for natural language processing: a survey[J]. Sci China Technol Sci, 2020, 63: 1872-1897.
[61]
Grechishnikova D. Transformer neural network for protein-specific de novo drug generation as a machine translation problem[J]. Sci Rep, 2021, 11: 321.
[62]
Blaschke T, Arus-Pous J, Chen H, et al. REINVENT 2.0: an AI tool for de novo drug design[J]. J Chem Inf Model, 2020, 60: 5918-5922.
63
Yang L, Zhang ZL, Song Y, et al. Diffusion models: a comprehensive survey of methods and applications [J]. arXiv, 2022: 2209.00796.
[64]
Ho J, Saharia C, Chan W, et al. Cascaded diffusion models for high fidelity image generation[J]. J Mach Learn Res, 2022, 23: 1-33
65
Li X, Thickstun J, Gulrajani I, et al. Diffusion-lm improves controllable text generation [J]. arXiv, 2022: 2205.14217.
[66]
Evans RM, Mangelsdorf DJ. Nuclear receptors, RXR, and the big bang[J]. Cell, 2014, 157: 255-266.
[67]
Lagarde N, Ben Nasr N, Jeremie A, et al. NRLiSt BDB, the manually curated nuclear receptors ligands and structures benchmarking database[J]. J Med Chem, 2014, 57: 3117-3125.
[68]
Merk D, Friedrich L, Grisoni F, et al. De novo design of bioactive small molecules by artificial intelligence[J]. Mol Inform, 2018, 37: 1700153.
[69]
Reker D, Rodrigues T, Schneider P, et al. Identifying the macromolecular targets of de novo-designed chemical entities through self-organizing map consensus[J]. Proc Natl Acad Sci, 2014, 111: 4067-4072.
[70]
Marder SR, Wirshing WC, Van Putten T. Drug treatment of schizophrenia: overview of recent research[J]. Schizophr Res, 1991, 4: 81-90.
[71]
Tan XQ, Jiang XR, He Y, et al. Automated design and optimization of multitarget schizophrenia drug candidates by deep learning[J]. Eur J Med Chem, 2020, 204: 112572.
[72]
Roskoski Jr R. Properties of FDA-approved small molecule protein kinase inhibitors: a 2023 update[J]. Pharmacol Res, 2022, 187: 106552
[73]
Roskoski Jr R. Classification of small molecule protein kinase inhibitors based upon the structures of their drug-enzyme complexes[J]. Pharmacol Res, 2016, 103: 26-48.
[74]
Putin E, Asadulaev A, Vanhaelen Q, et al. Adversarial threshold neural computer for molecular de novo design[J]. Mol Pharm, 2018, 15: 4386-4397.
[75]
Walters WP, Murcko M. Assessing the impact of generative AI on medicinal chemistry[J]. Nat Biotechnol, 2020, 38: 143-145.
[76]
Tan X, Li C, Yang R, et al. Discovery of pyrazolo[3, 4-d]pyridazinone derivatives as selective DDR1 inhibitors via deep learning based design, synthesis, and biological evaluation[J]. J Med Chem, 2022, 65: 103-119.
[77]
Li YS, Zhang LT, Wang YF, et al. Generative deep learning enables the discovery of a potent and selective RIPK1 inhibitor[J]. Nat Commun, 2022, 13: 6891.
[78]
Yang YX, Zhang RK, Li ZJ, et al. Discovery of highly potent, selective, and orally efficacious p300/CBP histone acetyltransferases inhibitors[J]. J Med Chem, 2020, 63: 1337-1360.
[79]
Wang ZQ, Ran T, Xu F, et al. Deep learning-driven scaffold hopping in the discovery of Akt kinase inhibitors[J]. Chem Commun, 2021, 57: 10588-10591.
[80]
Song SK, Tang HT, Ran T, et al. Application of deep generative model for design of pyrrolo[2, 3-d] pyrimidine derivatives as new selective TANK binding kinase 1 (TBK1) inhibitors[J]. Eur J Med Chem, 2022, 247: 115034
[81]
Liu RZ, Song JK, Liu AL, et al. Progress on the application of artificial intelligence technology in ligand-based and receptor structure-based drug screening[J]. Acta Pharm Sin (药学学报), 2021, 56: 2136-2145 http://www.yxxb.com.cn/aps/cn/article/doi/10.16438/j.0513-4870.2021-0052
[82]
Arús-Pous J, Blaschke T, Ulander S, et al. Exploring the GDB-13 chemical space using deep generative models[J]. J Cheminform, 2019, 11: 20.
[83]
Pierce AC, Rao G, Bemis GW. BREED:  generating novel inhibitors through hybridization of known ligands. Application to CDK2, P38, and HIV protease[J]. J Med Chem, 2004, 47: 2768-2775.
[84]
Schneider G, Clark DE. Automated de novo drug design: are we nearly there yet?[J]. Angew Chem Int Ed, 2019, 58: 10792-10803.
[85]
Rifaioglu AS, Atas H, Martin MJ, et al. Recent applications of deep learning and machine intelligence on in silico drug discovery: methods, tools and databases[J]. Brief Bioinform, 2019, 20: 1878-1912.
[86]
Irwin JJ, Tang KG, Young J, et al. ZINC20—a free ultralarge-scale chemical database for ligand discovery[J]. J Chem Inf Model, 2020, 60: 6065-6073.
[87]
Mendez D, Gaulton A, Bento AP, et al. ChEMBL: towards direct deposition of bioassay data[J]. Nucleic Acids Res, 2019, 47: D930-D940.
[88]
Blum LC, Reymond JL. 970 million druglike small molecules for virtual screening in the chemical universe database GDB-13[J]. J Am Chem Soc, 2009, 131: 8732-8733.
[89]
Moret M, Friedrich L, Grisoni F, et al. Generative molecular design in low data regimes[J]. Nat Mach Intell, 2020, 2: 171-180.
[90]
Brown N, Fiscato M, Segler MHS, et al. GuacaMol: benchmarking models for de novo molecular design[J]. J Chem Inf Model, 2019, 59: 1096-1108.
[91]
Pagadala NS, Syed K, Tuszynski J. Software for molecular docking: a review[J]. Biophys Rev, 2017, 9: 91-102.
[92]
Wallach I, Heifets A. Most ligand-based classification benchmarks reward memorization rather than generalization[J]. J Chem Inf Model, 2018, 58: 916-932.
[93]
Chen HM, Engkvist O, Wang YH, et al. The rise of deep learning in drug discovery[J]. Drug Discov Today, 2018, 23: 1241-1250.
2023年第58卷第9期
PDF下载
556
251
引用本文
BibTeX
文章信息
doi: 10.16438/j.0513-4870.2022-1453
  • 接收时间:2023-01-03
  • 首发时间:2025-11-21
  • 出版时间:2023-09-12
补充材料
相关文章
文章信息
作者
出版历史
  • 收稿日期:2023-01-03
  • 修回日期:2023-02-13
基金
中国医学科学院医学与健康科技创新工程(2021-I2M-1-069)
国家重大新药创制科技重大专项(2018ZX09711001-012-003)
作者信息
    中国医学科学院、北京协和医学院药物研究所, 天然药物活性物质与功能国家重点实验室, 北京 100050

通讯作者:

参考文献
分享链接
https://castjournals.cast.org.cn/joweb/yxxb/CN/10.16438/j.0513-4870.2022-1453
分享至
全文二维码

扫描看全文

引用本文
BibTeX
本文的引用情况
2种不同金属材料的力学参数

Family
属数
Number of
genus
种数
Number of
species
占总种数比例
Percentage of
total species (%)

Genus
种数
Number of
species
占总种数比例
Percentage of total
species (%)
鹅膏菌科Amanitaceae 2 11 5.26 鹅膏菌属 Amanita 10 4.78
小菇科 Mycenaceae 2 12 5.74 丝盖伞属 Inocybe 5 2.39
多孔菌科 Polyporaceae 8 14 6.70 蜡蘑属 Laccaria 5 2.39
红菇科 Russulaceae 3 23 11.00 小皮伞属 Marasmius 6 2.87
小菇属 Mycena 11 5.26
光柄菇属 Pluteus 5 2.39
红菇属 Russula 17 8.13
栓菌属 Trametes 5 2.39
关闭全屏