DrugPilot: LLM-based Parameterized Reasoning Agent for Drug Discovery

DrugPilot: LLM-based Parameterized Reasoning Agent for Drug Discovery | 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 DrugPilot: LLM-based Parameterized Reasoning Agent for Drug Discovery Wenbin Hu, Kun Li, Zhennan Wu, Shoupeng Wang, Jia Wu, Bo Du, Xiangyu Wang, and 1 more This is a preprint; it has not been peer reviewed by a journal. https://doi.org/ 10.21203/rs.3.rs-7489358/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 Large language models (LLMs) integrated with autonomous agents hold significant potential for advancing scientific discovery through automated reasoning and task execution. However, applying LLM agents to drug discovery is still constrained by challenges such as large-scale multimodal data processing, limited task automation, and poor support for domain-specific tools. To overcome these limitations, we introduce DrugPilot, a LLM-based agent system with a parameterized reasoning architecture designed for end-to-end scientific workflows in drug discovery. DrugPilot enables multi-stage research processes by integrating structured tool use with a novel parameterized memory pool. The memory pool converts heterogeneous data from both public sources and user-defined inputs into standardized representations. This design supports efficient multi-turn conversation, reduces information loss during data exchange, and enhances complex scientific decision-making. To support training and benchmarking, we construct a drug instruction dataset covering eight core drug discovery tasks. Under the Berkeley function-calling benchmark, DrugPilot significantly outperforms state-of-the-art agents such as ReAct and LoT, achieving task completion rates of 98.0%, 93.5%, and 64.0% for simple, multi-tool, and multi-turn categories, respectively. These results highlight DrugPilot's strong potential as a generalizable agent framework for automated, interactive, and data-driven reasoning across computational science applications. Biological sciences/Computational biology and bioinformatics Biological sciences/Drug discovery Biological sciences/Molecular biology large language model agent tool calling parameterized reasoning drug discovery Full Text Additional Declarations There is NO Competing Interest. Supplementary Files sm.pdf Supplementary 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-7489358","acceptedTermsAndConditions":true,"allowDirectSubmit":true,"archivedVersions":[],"articleType":"Article","associatedPublications":[],"authors":[{"id":520907666,"identity":"2120c47c-d8bc-42f4-9913-0ae00db8b1c9","order_by":0,"name":"Wenbin Hu","email":"data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAAZAAAAAyAQMAAABI0h/eAAAABlBMVEX///8AAABVwtN+AAAACXBIWXMAAA7EAAAOxAGVKw4bAAAAnUlEQVRIiWNgGAWjYBACAwYexgeMDSBmAvFamA1I1sImQZoWc4ncY9W8Ow4z8LPnGDD83EGEFssZeWm3ec8cZpDseWPA2HuGGIfdzjG7ndt2mMHgRo4BM2MbkVqKQVrsSdLCDLZFgmgt998lS/9tS+eROPOs4GAvUVrOnD34cWabtRx/e/LGBz+J0QIDPCDiAAkaRsEoGAWjYBTgAwAbjjVDYdGGpAAAAABJRU5ErkJggg==","orcid":"","institution":"Wuhan University","correspondingAuthor":true,"prefix":"","firstName":"Wenbin","middleName":"","lastName":"Hu","suffix":""},{"id":520907667,"identity":"042381df-4aaf-4be5-984a-e7893aaee586","order_by":1,"name":"Kun Li","email":"","orcid":"https://orcid.org/0000-0002-7002-234X","institution":"School of Computer Science, Wuhan University","correspondingAuthor":false,"prefix":"","firstName":"Kun","middleName":"","lastName":"Li","suffix":""},{"id":520907668,"identity":"c70744df-d06c-4ed7-82d2-a2a2eed0f7d1","order_by":2,"name":"Zhennan Wu","email":"","orcid":"","institution":"School of Computer Science, Wuhan University","correspondingAuthor":false,"prefix":"","firstName":"Zhennan","middleName":"","lastName":"Wu","suffix":""},{"id":520907669,"identity":"50a64688-db6c-4a59-b72d-a4ec14c85f33","order_by":3,"name":"Shoupeng Wang","email":"","orcid":"","institution":"School of Mathematics and Statistics, Wuhan University","correspondingAuthor":false,"prefix":"","firstName":"Shoupeng","middleName":"","lastName":"Wang","suffix":""},{"id":520907670,"identity":"d6d49ae6-00a9-4afa-9e90-81113fd3d007","order_by":4,"name":"Jia Wu","email":"","orcid":"","institution":"School of Computing, Macquarie University, Sydney, Australia","correspondingAuthor":false,"prefix":"","firstName":"Jia","middleName":"","lastName":"Wu","suffix":""},{"id":520907671,"identity":"51f39808-d145-4422-832d-1b560a0024e9","order_by":5,"name":"Bo Du","email":"","orcid":"","institution":"School of Computer Science, Wuhan University","correspondingAuthor":false,"prefix":"","firstName":"Bo","middleName":"","lastName":"Du","suffix":""},{"id":520907672,"identity":"f13d3d85-c8c9-43cc-ab1a-1872d9cba46e","order_by":6,"name":"Xiangyu Wang","email":"","orcid":"","institution":"Huazhong University of Science and Technology","correspondingAuthor":false,"prefix":"","firstName":"Xiangyu","middleName":"","lastName":"Wang","suffix":""},{"id":520907673,"identity":"50900c83-5483-44bb-9534-9c4d4d949f94","order_by":7,"name":"Shirui Pan","email":"","orcid":"https://orcid.org/0000-0003-0794-527X","institution":"Griffith University","correspondingAuthor":false,"prefix":"","firstName":"Shirui","middleName":"","lastName":"Pan","suffix":""}],"badges":[],"createdAt":"2025-08-29 14:11:51","currentVersionCode":1,"declarations":"","doi":"10.21203/rs.3.rs-7489358/v1","doiUrl":"https://doi.org/10.21203/rs.3.rs-7489358/v1","draftVersion":[],"editorialEvents":[],"editorialNote":"","failedWorkflow":false,"files":[{"id":98430398,"identity":"3585e95a-9487-4e17-8a15-cfd1df6e5724","added_by":"auto","created_at":"2025-12-17 16:45:18","extension":"pdf","order_by":1,"title":"","display":"","copyAsset":false,"role":"manuscript-pdf","size":2848247,"visible":true,"origin":"","legend":"Article File","description":"","filename":"main.pdf","url":"https://assets-eu.researchsquare.com/files/rs-7489358/v1_covered_1837dcfe-ca4a-445a-93b9-3afa6ef696d3.pdf"},{"id":92375292,"identity":"9c51c465-03bb-420a-9c97-35a0e9729d0c","added_by":"auto","created_at":"2025-09-29 04:27:12","extension":"pdf","order_by":1,"title":"","display":"","copyAsset":false,"role":"supplement","size":576176,"visible":true,"origin":"","legend":"Supplementary Information","description":"","filename":"sm.pdf","url":"https://assets-eu.researchsquare.com/files/rs-7489358/v1/ae6644337ce5a2b779e9bd27.pdf"}],"financialInterests":"There is \u003cb\u003eNO\u003c/b\u003e Competing Interest.","formattedTitle":"DrugPilot: LLM-based Parameterized Reasoning Agent for Drug Discovery","fulltext":[],"fulltextSource":"","fullText":"","funders":[],"hasAdminPriorityOnWorkflow":false,"hasManuscriptDocX":false,"hasOptedInToPreprint":true,"hasPassedJournalQc":"","hasAnyPriority":true,"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":"large language model, agent, tool calling, parameterized reasoning, drug discovery","lastPublishedDoi":"10.21203/rs.3.rs-7489358/v1","lastPublishedDoiUrl":"https://doi.org/10.21203/rs.3.rs-7489358/v1","license":{"name":"CC BY 4.0","url":"https://creativecommons.org/licenses/by/4.0/"},"manuscriptAbstract":"Large language models (LLMs) integrated with autonomous agents hold significant potential for advancing scientific discovery through automated reasoning and task execution. However, applying LLM agents to drug discovery is still constrained by challenges such as large-scale multimodal data processing, limited task automation, and poor support for domain-specific tools. To overcome these limitations, we introduce DrugPilot, a LLM-based agent system with a parameterized reasoning architecture designed for end-to-end scientific workflows in drug discovery. DrugPilot enables multi-stage research processes by integrating structured tool use with a novel parameterized memory pool. The memory pool converts heterogeneous data from both public sources and user-defined inputs into standardized representations. This design supports efficient multi-turn conversation, reduces information loss during data exchange, and enhances complex scientific decision-making. To support training and benchmarking, we construct a drug instruction dataset covering eight core drug discovery tasks. Under the Berkeley function-calling benchmark, DrugPilot significantly outperforms state-of-the-art agents such as ReAct and LoT, achieving task completion rates of 98.0%, 93.5%, and 64.0% for simple, multi-tool, and multi-turn categories, respectively. These results highlight DrugPilot's strong potential as a generalizable agent framework for automated, interactive, and data-driven reasoning across computational science applications.","manuscriptTitle":"DrugPilot: LLM-based Parameterized Reasoning Agent for Drug Discovery","msid":"","msnumber":"","nonDraftVersions":[{"code":1,"date":"2025-09-29 04:27:07","doi":"10.21203/rs.3.rs-7489358/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":"6a3ef3a7-1560-4752-a6d9-8849cc293f36","owner":[],"postedDate":"September 29th, 2025","published":true,"recentEditorialEvents":[],"rejectedJournal":[],"revision":"","amendment":"","status":"posted","subjectAreas":[{"id":55370892,"name":"Biological sciences/Computational biology and bioinformatics"},{"id":55370893,"name":"Biological sciences/Drug discovery"},{"id":55370894,"name":"Biological sciences/Molecular biology"}],"tags":[],"updatedAt":"2026-01-15T21:05:13+00:00","versionOfRecord":[],"versionCreatedAt":"2025-09-29 04:27:07","video":"","vorDoi":"","vorDoiUrl":"","workflowStages":[]},"version":"v1","identity":"rs-7489358","journalConfig":"researchsquare"},"__N_SSP":true},"page":"/article/[identity]/[[...version]]","query":{"redirect":"/article/rs-7489358","identity":"rs-7489358","version":["v1"]},"buildId":"8U1c8b4HqxoKbykW_rLl7","isFallback":false,"isExperimentalCompile":false,"dynamicIds":[84888],"gssp":true,"scriptLoader":[]}