LM-PROTAC: a language model-driven PROTAC generation pipeline with dual constraints of structure and property

LM-PROTAC: a language model-driven PROTAC generation pipeline with dual constraints of structure and property | Research Square window.SnipcartSettings = { analytics: { enabled: false } }; (function() { var accessVector = localStorage.getItem('access_vector') || ''; window.dataLayer = window.dataLayer || []; if (accessVector) { window.dataLayer.push({ user: { profile: { profileInfo: { snid: accessVector } } } }); } })(); (function(w,d,s,l,i){w[l]=w[l]||[];w[l].push({'gtm.start':new Date().getTime(),event:'gtm.js'});var f=d.getElementsByTagName(s)[0],j=d.createElement(s),dl=l!='dataLayer'?'&l='+l:'';j.async=true;j.src='https://www.googletagmanager.com/gtm.js?id='+i+dl;f.parentNode.insertBefore(j,f);})(window,document,'script','dataLayer','GTM-K279D39R'); Browse Preprints In Review Journals COVID-19 Preprints AJE Video Bytes Research Tools Research Promotion AJE Professional Editing AJE Rubriq About Preprint Platform In Review Editorial Policies Our Team Advisory Board Help Center Sign In Submit a Preprint Cite Share Download PDF Article LM-PROTAC: a language model-driven PROTAC generation pipeline with dual constraints of structure and property Li Wang, Shao Jinsong, Qineng Gong, Zeyu Yin, Yu Chen, Yajie Hao, and 5 more This is a preprint; it has not been peer reviewed by a journal. https://doi.org/ 10.21203/rs.3.rs-6356959/v1 This work is licensed under a CC BY 4.0 License Status: Posted Version 1 posted You are reading this latest preprint version Abstract The imperfect modeling of ternary complexes has limited the application of computer-aided drug discovery tools in PROTAC research and development. In this study, a language model for PROTAC molecule design pipeline named LM-PROTAC was developed, which stands for language model-driven Proteolysis Targeting Chimera, by embedding a transformer-based generative model with dual constraints on structure and properties. This study started with the idea of segmentation and representation of molecules and protein. Firstly, a language model-driven affinity model for protein compounds to screen molecular fragments with high affinity for the target protein. Secondly, structural and physicochemical properties of these fragments were constrained during the generation process to meet specific scenario requirements. Finally, a two-round screening was performed on the preliminary generated molecules using a multidimensional property prediction model. This process identified a batch of PROTAC molecules capable of degrading disease-relevant target proteins. These molecules were subsequently validated through in vitro experiments, thus providing a complete solution for language model-driven PROTAC drug generation. Taking Wnt3a, a key tumor-related target, as a POI of degradation, the LM-PROTAC pipeline successfully generated effective PROTAC molecules. The molecular distribution experiments demonstrated the high similarity of the generated molecules to the original dataset, validating the generative model’s effectiveness in accurately defining chemical space. Molecular dynamics simulations confirmed the stable interactions between the PROTAC molecules and target proteins, while protein degradation experiments verified the efficacy of the generated PROTAC molecules in degrading target proteins. The entire LM-PROTAC pipeline is reusable and can generate degraders for other target proteins within 50 days, significantly improving the efficiency of drug discovery for undruggable targets. Biological sciences/Chemical biology/Computational chemistry Biological sciences/Chemical biology/Cheminformatics Biological sciences/Structural biology/Molecular modelling Generative models Artificial intelligence Molecular generation Targeted protein degradation Drug discovery pipeline Full Text Additional Declarations There is NO Competing Interest. Supplementary Files SupportInformation.docx Support information Cite Share Download PDF Status: Posted Version 1 posted You are reading this latest preprint version Research Square lets you share your work early, gain feedback from the community, and start making changes to your manuscript prior to peer review in a journal. As a division of Research Square Company, we’re committed to making research communication faster, fairer, and more useful. We do this by developing innovative software and high quality services for the global research community. Our growing team is made up of researchers and industry professionals working together to solve the most critical problems facing scientific publishing. Also discoverable on Platform About Our Team In Review Editorial Policies Advisory Board Help Center Resources Author Services Accessibility API Access RSS feed Manage Cookie Preferences © Research Square 2026 | ISSN 2693-5015 (online) Privacy Policy Terms of Service Do Not Sell My Personal Information {"props":{"pageProps":{"initialData":{"identity":"rs-6356959","acceptedTermsAndConditions":true,"allowDirectSubmit":true,"archivedVersions":[],"articleType":"Article","associatedPublications":[],"authors":[{"id":437844813,"identity":"4cf1a656-1b80-4d76-a377-c0cf71ed1747","order_by":0,"name":"Li Wang","email":"data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAAZAAAAAyAQMAAABI0h/eAAAABlBMVEX///8AAABVwtN+AAAACXBIWXMAAA7EAAAOxAGVKw4bAAAAwUlEQVRIiWNgGAWjYLCCBwY2EAYP0VoSDNIgqknQwnCYBC0Gx88efpFQcD7PXiKB8cHbNgZ5c4JazuSlWSQY3C7mkUhgNpzbxmC4s4GAFrMDOWYGQC2JPRIJbNK8bUB/HSCk5fwbkJZzIC3sv4nTciPH+AFQGdgWZqK02N94YwZUlpzYc+Zhs+SccxKGGwhpkezPMf7w4Y9dYnt78sEPb8ps5AnaAgRsEhCasQFISBBWDwTMH4hSNgpGwSgYBSMXAADOtj8zcW2ZJgAAAABJRU5ErkJggg==","orcid":"https://orcid.org/0000-0003-2838-9807","institution":"Nantong University","correspondingAuthor":true,"prefix":"","firstName":"Li","middleName":"","lastName":"Wang","suffix":""},{"id":437844814,"identity":"a954ff11-2fc9-4bd7-8cd7-084e51000f5a","order_by":1,"name":"Shao Jinsong","email":"","orcid":"https://orcid.org/0000-0001-5883-1016","institution":"Nantong University","correspondingAuthor":false,"prefix":"","firstName":"Shao","middleName":"","lastName":"Jinsong","suffix":""},{"id":437844815,"identity":"1b778ed4-67bd-4e6b-bec7-ab55cdd4aeb9","order_by":2,"name":"Qineng Gong","email":"","orcid":"","institution":"","correspondingAuthor":false,"prefix":"","firstName":"Qineng","middleName":"","lastName":"Gong","suffix":""},{"id":437844816,"identity":"01b223a6-e77d-4902-b0f2-6f9705b58bdd","order_by":3,"name":"Zeyu Yin","email":"","orcid":"","institution":"","correspondingAuthor":false,"prefix":"","firstName":"Zeyu","middleName":"","lastName":"Yin","suffix":""},{"id":437844817,"identity":"ffc9eb53-8f48-42f2-b669-82fb34441d2d","order_by":4,"name":"Yu Chen","email":"","orcid":"","institution":"","correspondingAuthor":false,"prefix":"","firstName":"Yu","middleName":"","lastName":"Chen","suffix":""},{"id":437844818,"identity":"503ef034-8cb9-4e14-ad93-77330d7a226b","order_by":5,"name":"Yajie Hao","email":"","orcid":"https://orcid.org/0000-0002-0811-0445","institution":"","correspondingAuthor":false,"prefix":"","firstName":"Yajie","middleName":"","lastName":"Hao","suffix":""},{"id":437844819,"identity":"a682dbbf-493d-40c5-87f1-43e8aa72d482","order_by":6,"name":"Lei Zhang","email":"","orcid":"","institution":"","correspondingAuthor":false,"prefix":"","firstName":"Lei","middleName":"","lastName":"Zhang","suffix":""},{"id":437844820,"identity":"a8565727-c22a-45ad-ab59-2de4bf304c65","order_by":7,"name":"Linlin Jiang","email":"","orcid":"","institution":"","correspondingAuthor":false,"prefix":"","firstName":"Linlin","middleName":"","lastName":"Jiang","suffix":""},{"id":437844821,"identity":"3bb2d4c5-425d-40e0-b2ba-a0c156fa5f9d","order_by":8,"name":"Min Yao","email":"","orcid":"","institution":"","correspondingAuthor":false,"prefix":"","firstName":"Min","middleName":"","lastName":"Yao","suffix":""},{"id":437844822,"identity":"a4cb2274-383f-4987-a950-362059a29f93","order_by":9,"name":"Jinlong Li","email":"","orcid":"","institution":"","correspondingAuthor":false,"prefix":"","firstName":"Jinlong","middleName":"","lastName":"Li","suffix":""},{"id":437844823,"identity":"fcf61015-6a2a-4d56-978b-6539b8cb9925","order_by":10,"name":"Fubo Wang","email":"","orcid":"","institution":"","correspondingAuthor":false,"prefix":"","firstName":"Fubo","middleName":"","lastName":"Wang","suffix":""}],"badges":[],"createdAt":"2025-04-02 03:05:30","currentVersionCode":1,"declarations":"","doi":"10.21203/rs.3.rs-6356959/v1","doiUrl":"https://doi.org/10.21203/rs.3.rs-6356959/v1","draftVersion":[],"editorialEvents":[],"editorialNote":"","failedWorkflow":false,"files":[{"id":85320358,"identity":"ea4932a9-7543-4679-80be-688d877319e5","added_by":"auto","created_at":"2025-06-24 15:20:00","extension":"pdf","order_by":1,"title":"","display":"","copyAsset":false,"role":"manuscript-pdf","size":1296091,"visible":true,"origin":"","legend":"","description":"","filename":"Manuscript.pdf","url":"https://assets-eu.researchsquare.com/files/rs-6356959/v1_covered_e8541ceb-bfd2-4191-ad1e-b1be5645349d.pdf"},{"id":80023016,"identity":"1f95854a-5b77-4fe4-910a-c45a3374ff04","added_by":"auto","created_at":"2025-04-07 05:29:26","extension":"docx","order_by":1,"title":"","display":"","copyAsset":false,"role":"supplement","size":2607520,"visible":true,"origin":"","legend":"Support information","description":"","filename":"SupportInformation.docx","url":"https://assets-eu.researchsquare.com/files/rs-6356959/v1/3e0de08cc96be05e59c89be4.docx"}],"financialInterests":"There is \u003cb\u003eNO\u003c/b\u003e Competing Interest.","formattedTitle":"LM-PROTAC: a language model-driven PROTAC generation pipeline with dual constraints of structure and property","fulltext":[],"fulltextSource":"","fullText":"","funders":[],"hasAdminPriorityOnWorkflow":false,"hasManuscriptDocX":false,"hasOptedInToPreprint":true,"hasPassedJournalQc":"","hasAnyPriority":false,"hideJournal":true,"highlight":"","institution":"","isAcceptedByJournal":false,"isAuthorSuppliedPdf":true,"isDeskRejected":"","isHiddenFromSearch":false,"isInQc":false,"isInWorkflow":false,"isPdf":true,"isPdfUpToDate":true,"isWithdrawnOrRetracted":false,"journal":{"display":true,"email":"[email protected]","identity":"researchsquare","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":true,"externalIdentity":"","sideBox":"","snPcode":"","submissionUrl":"/submission","title":"Research Square","twitterHandle":"researchsquare","acdcEnabled":true,"dfaEnabled":false,"editorialSystem":"","reportingPortfolio":"","inReviewEnabled":false,"inReviewRevisionsEnabled":true},"keywords":"Generative models, Artificial intelligence, Molecular generation, Targeted protein degradation, Drug discovery pipeline","lastPublishedDoi":"10.21203/rs.3.rs-6356959/v1","lastPublishedDoiUrl":"https://doi.org/10.21203/rs.3.rs-6356959/v1","license":{"name":"CC BY 4.0","url":"https://creativecommons.org/licenses/by/4.0/"},"manuscriptAbstract":"\u003cp\u003eThe imperfect modeling of ternary complexes has limited the application of computer-aided drug discovery tools in PROTAC research and development. In this study, a language model for PROTAC molecule design pipeline named LM-PROTAC was developed, which stands for language model-driven Proteolysis Targeting Chimera, by embedding a transformer-based generative model with dual constraints on structure and properties. This study started with the idea of segmentation and representation of molecules and protein. Firstly, a language model-driven affinity model for protein compounds to screen molecular fragments with high affinity for the target protein. Secondly, structural and physicochemical properties of these fragments were constrained during the generation process to meet specific scenario requirements. Finally, a two-round screening was performed on the preliminary generated molecules using a multidimensional property prediction model. This process identified a batch of PROTAC molecules capable of degrading disease-relevant target proteins. These molecules were subsequently validated through in vitro experiments, thus providing a complete solution for language model-driven PROTAC drug generation. Taking Wnt3a, a key tumor-related target, as a POI of degradation, the LM-PROTAC pipeline successfully generated effective PROTAC molecules. The molecular distribution experiments demonstrated the high similarity of the generated molecules to the original dataset, validating the generative model\u0026rsquo;s effectiveness in accurately defining chemical space. Molecular dynamics simulations confirmed the stable interactions between the PROTAC molecules and target proteins, while protein degradation experiments verified the efficacy of the generated PROTAC molecules in degrading target proteins. The entire LM-PROTAC pipeline is reusable and can generate degraders for other target proteins within 50 days, significantly improving the efficiency of drug discovery for undruggable targets.\u003c/p\u003e","manuscriptTitle":"LM-PROTAC: a language model-driven PROTAC generation pipeline with dual constraints of structure and property","msid":"","msnumber":"","nonDraftVersions":[{"code":1,"date":"2025-04-07 05:05:21","doi":"10.21203/rs.3.rs-6356959/v1","editorialEvents":[{"type":"communityComments","content":0}],"status":"published","journal":{"display":true,"email":"[email protected]","identity":"researchsquare","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":true,"externalIdentity":"","sideBox":"","snPcode":"","submissionUrl":"/submission","title":"Research Square","twitterHandle":"researchsquare","acdcEnabled":true,"dfaEnabled":false,"editorialSystem":"","reportingPortfolio":"","inReviewEnabled":false,"inReviewRevisionsEnabled":true}}],"origin":"","ownerIdentity":"7badd2d9-cb2d-49c7-ad07-23665c0eccfe","owner":[],"postedDate":"April 7th, 2025","published":true,"recentEditorialEvents":[],"rejectedJournal":[],"revision":"","amendment":"","status":"posted","subjectAreas":[{"id":46622603,"name":"Biological sciences/Chemical biology/Computational chemistry"},{"id":46622604,"name":"Biological sciences/Chemical biology/Cheminformatics"},{"id":46622605,"name":"Biological sciences/Structural biology/Molecular modelling"}],"tags":[],"updatedAt":"2025-06-24T15:11:51+00:00","versionOfRecord":[],"versionCreatedAt":"2025-04-07 05:05:21","video":"","vorDoi":"","vorDoiUrl":"","workflowStages":[]},"version":"v1","identity":"rs-6356959","journalConfig":"researchsquare"},"__N_SSP":true},"page":"/article/[identity]/[[...version]]","query":{"redirect":"/article/rs-6356959","identity":"rs-6356959","version":["v1"]},"buildId":"8U1c8b4HqxoKbykW_rLl7","isFallback":false,"isExperimentalCompile":false,"dynamicIds":[84888],"gssp":true,"scriptLoader":[]}