Benchmarking Molecular Representations for Aqueous Solubility Prediction: The Impact of Inductive Bias and Scaffold Splitting in Low-Data Regimes

Benchmarking Molecular Representations for Aqueous Solubility Prediction: The Impact of Inductive Bias and Scaffold Splitting in Low-Data Regimes | 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 Research Article Benchmarking Molecular Representations for Aqueous Solubility Prediction: The Impact of Inductive Bias and Scaffold Splitting in Low-Data Regimes Mudassir Ur Rahman This is a preprint; it has not been peer reviewed by a journal. https://doi.org/ 10.21203/rs.3.rs-9059650/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 Accurate prediction of aqueous solubility (logS) is a critical bottleneck in the early stages of drug discovery and formulation. While Graph Neural Networks (GNNs) have emerged as state-of-the-art architectures for molecular property prediction, their efficacy compared to classical feature engineer- ing remains contested in low-data regimes. In this study, we perform a rigorous comparative analysis of three molecular representation strategies—explicit physicochemical descriptors, high-dimensional Morgan fingerprints, and end-to-end graph embeddings—evaluated on the ESOL dataset ( N = 1 , 128). To simulate realistic prospective evaluation, we employ a Murcko scaffold split, ensuring that the test set contains novel chemotypes distinct from the training distribution. Our results demonstrate that a Multi-Layer Perceptron (MLP) trained on domain-specific descriptors (e.g., LogP, Molecular Weight) achieves superior per- formance (MAE = 0 . 73, R 2 = 0 . 75), significantly outperforming both Morgan fingerprints ( R 2 = 0 . 14) and Graph Convolutional Networks ( R 2 = 0 . 54). This suggests that for small-scale datasets, the inductive bias provided by explicit physical features outweighs the representation learning capabilities of GNNs. Furthermore, we implement a Deep Ensemble framework to quantify predictive uncertainty. We find a strong correlation between ensemble variance and prediction error, validating the use of uncertainty estimation as a reliability filter for out-of-domain screening. These findings advocate for a ”physics-first” approach when applying deep learning to small, sparse chemical datasets. Aqueous Solubility Graph Neural Networks QSAR Scaffold Splitting Uncertainty Quantification Deep Learning Full Text Additional Declarations No competing interests reported. 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-9059650","acceptedTermsAndConditions":true,"allowDirectSubmit":true,"archivedVersions":[],"articleType":"Research Article","associatedPublications":[],"authors":[{"id":610153374,"identity":"736078c1-9490-4c36-b9db-e171f275944d","order_by":0,"name":"Mudassir Ur Rahman","email":"data:image/png;base64,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","orcid":"","institution":"GDC Lund Khwar, Abdul Wali Khan University Mardan","correspondingAuthor":true,"prefix":"","firstName":"Mudassir","middleName":"Ur","lastName":"Rahman","suffix":""}],"badges":[],"createdAt":"2026-03-07 15:53:52","currentVersionCode":1,"declarations":"","doi":"10.21203/rs.3.rs-9059650/v1","doiUrl":"https://doi.org/10.21203/rs.3.rs-9059650/v1","draftVersion":[],"editorialEvents":[],"editorialNote":"","failedWorkflow":false,"files":[{"id":107272963,"identity":"68a5c6be-f1cf-43e2-9871-775959de95cb","added_by":"auto","created_at":"2026-04-19 19:54:02","extension":"pdf","order_by":1,"title":"","display":"","copyAsset":false,"role":"manuscript-pdf","size":608953,"visible":true,"origin":"","legend":"","description":"","filename":"Chemistry11.pdf","url":"https://assets-eu.researchsquare.com/files/rs-9059650/v1_covered_f36a453e-7de6-46e8-8694-cceefdd816df.pdf"}],"financialInterests":"No competing interests reported.","formattedTitle":"Benchmarking Molecular Representations for Aqueous Solubility Prediction: The Impact of Inductive Bias and Scaffold Splitting in Low-Data Regimes","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":"Aqueous Solubility, Graph Neural Networks, QSAR, Scaffold Splitting, Uncertainty Quantification, Deep Learning","lastPublishedDoi":"10.21203/rs.3.rs-9059650/v1","lastPublishedDoiUrl":"https://doi.org/10.21203/rs.3.rs-9059650/v1","license":{"name":"CC BY 4.0","url":"https://creativecommons.org/licenses/by/4.0/"},"manuscriptAbstract":"\u003cp\u003eAccurate prediction of aqueous solubility (logS) is a critical bottleneck in the early stages of drug discovery and formulation. While Graph Neural Networks (GNNs) have emerged as state-of-the-art architectures for molecular property prediction, their efficacy compared to classical feature engineer- ing remains contested in low-data regimes. In this study, we perform a rigorous comparative analysis of three molecular representation strategies—explicit physicochemical descriptors, high-dimensional Morgan fingerprints, and end-to-end graph embeddings—evaluated on the ESOL dataset (\u003cem\u003eN \u003c/em\u003e= 1\u003cem\u003e, \u003c/em\u003e128). To simulate realistic prospective evaluation, we employ a Murcko scaffold split, ensuring that the test set contains novel chemotypes distinct from the training distribution. Our results demonstrate that a Multi-Layer Perceptron (MLP) trained on domain-specific\u003c/p\u003e\n\u003cp\u003edescriptors (e.g., LogP, Molecular Weight) achieves superior per- formance (MAE = 0\u003cem\u003e.\u003c/em\u003e73, \u003cem\u003eR\u003c/em\u003e\u003csup\u003e2\u003c/sup\u003e = 0\u003cem\u003e.\u003c/em\u003e75), significantly outperforming both Morgan fingerprints (\u003cem\u003eR\u003c/em\u003e\u003csup\u003e2\u003c/sup\u003e = 0\u003cem\u003e.\u003c/em\u003e14) and Graph Convolutional Networks (\u003cem\u003eR\u003c/em\u003e\u003csup\u003e2\u003c/sup\u003e = 0\u003cem\u003e.\u003c/em\u003e54). This suggests that for small-scale datasets, the inductive bias provided by explicit physical features outweighs the representation learning capabilities of GNNs. Furthermore, we implement a Deep Ensemble framework to quantify predictive uncertainty. We find a strong correlation between ensemble\u003c/p\u003e\n\u003cp\u003evariance and prediction error, validating the use of uncertainty estimation as a reliability filter for out-of-domain screening. These findings advocate for a ”physics-first” approach when applying deep learning to small, sparse chemical datasets.\u003c/p\u003e","manuscriptTitle":"Benchmarking Molecular Representations for Aqueous Solubility Prediction: The Impact of Inductive Bias and Scaffold Splitting in Low-Data Regimes","msid":"","msnumber":"","nonDraftVersions":[{"code":1,"date":"2026-03-23 18:31:17","doi":"10.21203/rs.3.rs-9059650/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":"bbed4421-8d11-4b56-96d8-6512f55d2f5e","owner":[],"postedDate":"March 23rd, 2026","published":true,"recentEditorialEvents":[],"rejectedJournal":[],"revision":"","amendment":"","status":"posted","subjectAreas":[],"tags":[],"updatedAt":"2026-04-19T19:53:39+00:00","versionOfRecord":[],"versionCreatedAt":"2026-03-23 18:31:17","video":"","vorDoi":"","vorDoiUrl":"","workflowStages":[]},"version":"v1","identity":"rs-9059650","journalConfig":"researchsquare"},"__N_SSP":true},"page":"/article/[identity]/[[...version]]","query":{"redirect":"/article/rs-9059650","identity":"rs-9059650","version":["v1"]},"buildId":"XKTyCvWXoU3ODBz1xrDgd","isFallback":false,"isExperimentalCompile":false,"dynamicIds":[84888],"gssp":true,"scriptLoader":[]}