Human-in-the-loop active learning for goal-oriented molecule generation

IF 7.1 2区 化学 Q1 CHEMISTRY, MULTIDISCIPLINARY
Yasmine Nahal, Janosch Menke, Julien Martinelli, Markus Heinonen, Mikhail Kabeshov, Jon Paul Janet, Eva Nittinger, Ola Engkvist, Samuel Kaski
{"title":"Human-in-the-loop active learning for goal-oriented molecule generation","authors":"Yasmine Nahal,&nbsp;Janosch Menke,&nbsp;Julien Martinelli,&nbsp;Markus Heinonen,&nbsp;Mikhail Kabeshov,&nbsp;Jon Paul Janet,&nbsp;Eva Nittinger,&nbsp;Ola Engkvist,&nbsp;Samuel Kaski","doi":"10.1186/s13321-024-00924-y","DOIUrl":null,"url":null,"abstract":"<p>Machine learning (ML) systems have enabled the modelling of quantitative structure–property relationships (QSPR) and structure-activity relationships (QSAR) using existing experimental data to predict target properties for new molecules. These property predictors hold significant potential in accelerating drug discovery by guiding generative artificial intelligence (AI) agents to explore desired chemical spaces. However, they often struggle to generalize due to the limited scope of the training data. When optimized by generative agents, this limitation can result in the generation of molecules with artificially high predicted probabilities of satisfying target properties, which subsequently fail experimental validation. To address this challenge, we propose an adaptive approach that integrates active learning (AL) and iterative feedback to refine property predictors, thereby improving the outcomes of their optimization by generative AI agents. Our method leverages the Expected Predictive Information Gain (EPIG) criterion to select additional molecules for evaluation by an oracle. This process aims to provide the greatest reduction in predictive uncertainty, enabling more accurate model evaluations of subsequently generated molecules. Recognizing the impracticality of immediate wet-lab or physics-based experiments due to time and logistical constraints, we propose leveraging human experts for their cost-effectiveness and domain knowledge to effectively augment property predictors, bridging gaps in the limited training data. Empirical evaluations through both simulated and real human-in-the-loop experiments demonstrate that our approach refines property predictors to better align with oracle assessments. Additionally, we observe improved accuracy of predicted properties as well as improved drug-likeness among the top-ranking generated molecules.</p><p>We present an adaptable framework that integrates AL and human expertise to refine property predictors for goal-oriented molecule generation. This approach is robust to noise in human feedback and ensures that navigating chemical space with human-refined predictors leverages human insights to identify molecules that not only satisfy predicted property profiles but also score highly on oracle models. Additionally, it prioritizes practical characteristics such as drug-likeness, synthetic accessibility, and a favorable balance between exploring diverse chemical space and exploiting similarity to existing training data.</p>","PeriodicalId":617,"journal":{"name":"Journal of Cheminformatics","volume":"16 1","pages":""},"PeriodicalIF":7.1000,"publicationDate":"2024-12-09","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://jcheminf.biomedcentral.com/counter/pdf/10.1186/s13321-024-00924-y","citationCount":"0","resultStr":null,"platform":"Semanticscholar","paperid":null,"PeriodicalName":"Journal of Cheminformatics","FirstCategoryId":"92","ListUrlMain":"https://link.springer.com/article/10.1186/s13321-024-00924-y","RegionNum":2,"RegionCategory":"化学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":null,"EPubDate":"","PubModel":"","JCR":"Q1","JCRName":"CHEMISTRY, MULTIDISCIPLINARY","Score":null,"Total":0}
引用次数: 0

Abstract

Machine learning (ML) systems have enabled the modelling of quantitative structure–property relationships (QSPR) and structure-activity relationships (QSAR) using existing experimental data to predict target properties for new molecules. These property predictors hold significant potential in accelerating drug discovery by guiding generative artificial intelligence (AI) agents to explore desired chemical spaces. However, they often struggle to generalize due to the limited scope of the training data. When optimized by generative agents, this limitation can result in the generation of molecules with artificially high predicted probabilities of satisfying target properties, which subsequently fail experimental validation. To address this challenge, we propose an adaptive approach that integrates active learning (AL) and iterative feedback to refine property predictors, thereby improving the outcomes of their optimization by generative AI agents. Our method leverages the Expected Predictive Information Gain (EPIG) criterion to select additional molecules for evaluation by an oracle. This process aims to provide the greatest reduction in predictive uncertainty, enabling more accurate model evaluations of subsequently generated molecules. Recognizing the impracticality of immediate wet-lab or physics-based experiments due to time and logistical constraints, we propose leveraging human experts for their cost-effectiveness and domain knowledge to effectively augment property predictors, bridging gaps in the limited training data. Empirical evaluations through both simulated and real human-in-the-loop experiments demonstrate that our approach refines property predictors to better align with oracle assessments. Additionally, we observe improved accuracy of predicted properties as well as improved drug-likeness among the top-ranking generated molecules.

We present an adaptable framework that integrates AL and human expertise to refine property predictors for goal-oriented molecule generation. This approach is robust to noise in human feedback and ensures that navigating chemical space with human-refined predictors leverages human insights to identify molecules that not only satisfy predicted property profiles but also score highly on oracle models. Additionally, it prioritizes practical characteristics such as drug-likeness, synthetic accessibility, and a favorable balance between exploring diverse chemical space and exploiting similarity to existing training data.

求助全文
约1分钟内获得全文 求助全文
来源期刊
Journal of Cheminformatics
Journal of Cheminformatics CHEMISTRY, MULTIDISCIPLINARY-COMPUTER SCIENCE, INFORMATION SYSTEMS
CiteScore
14.10
自引率
7.00%
发文量
82
审稿时长
3 months
期刊介绍: Journal of Cheminformatics is an open access journal publishing original peer-reviewed research in all aspects of cheminformatics and molecular modelling. Coverage includes, but is not limited to: chemical information systems, software and databases, and molecular modelling, chemical structure representations and their use in structure, substructure, and similarity searching of chemical substance and chemical reaction databases, computer and molecular graphics, computer-aided molecular design, expert systems, QSAR, and data mining techniques.
×
引用
GB/T 7714-2015
复制
MLA
复制
APA
复制
导出至
BibTeX EndNote RefMan NoteFirst NoteExpress
×
提示
您的信息不完整,为了账户安全,请先补充。
现在去补充
×
提示
您因"违规操作"
具体请查看互助需知
我知道了
×
提示
确定
请完成安全验证×
copy
已复制链接
快去分享给好友吧!
我知道了
右上角分享
点击右上角分享
0
联系我们:info@booksci.cn Book学术提供免费学术资源搜索服务,方便国内外学者检索中英文文献。致力于提供最便捷和优质的服务体验。 Copyright © 2023 布克学术 All rights reserved.
京ICP备2023020795号-1
ghs 京公网安备 11010802042870号
Book学术文献互助
Book学术文献互助群
群 号:481959085
Book学术官方微信