Multi-omics cancer subtyping via attention-guided variational representation learning and deep clustering

Multi-omics cancer subtyping via attention-guided variational representation learning and deep clustering | 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 Multi-omics cancer subtyping via attention-guided variational representation learning and deep clustering Xinning Liu, Li Han, Ling Kang, Lei Zhao, Quan Guo, Huadong Miao This is a preprint; it has not been peer reviewed by a journal. https://doi.org/ 10.21203/rs.3.rs-9259481/v1 This work is licensed under a CC BY 4.0 License Status: Under Review Version 1 posted 8 You are reading this latest preprint version Abstract Tumor heterogeneity is a major challenge for accurate molecular subtyping and precision oncology, as complex variations across and within tumors can lead to distinct biological behaviors, treatment responses, and clinical outcomes. To address this issue, we propose MAVDC, a multi-omics integration framework that combines unsupervised feature selection, variational representation learning, attention-based fusion, contrastive learning, and adaptive deep embedded clustering. MAVDC first ranks features within each omics modality to suppress noise and retain informative signals, and then determines the optimal number of clusters using a composite scoring strategy. Compact variational autoencoders are employed to learn modality-specific latent representations, which are further integrated through a multi-head attention mechanism. A progressive training strategy jointly optimizes reconstruction, variational regularization, clustering refinement, and cross-omics consistency. Experiments on TCGA BRCA and LUSC cohorts integrating RNA expression, DNA methylation, and copy number variation data identified four molecular subtypes in each cancer. These subtypes exhibited clear geometric separation, distinct functional and immune characteristics, and significant survival differences, outperforming representative single-omics and multi-omics baseline methods. MAVDC provides an effective tool for discovering clinically meaningful cancer subtypes. Biological sciences/Cancer Biological sciences/Computational biology and bioinformatics multi-omics integration cancer subtyping variational autoencoder attention-based fusion deep embedded clustering survival stratification Full Text Additional Declarations No competing interests reported. Cite Share Download PDF Status: Under Review Version 1 posted Reviewers agreed at journal 26 Apr, 2026 Reviewers agreed at journal 23 Apr, 2026 Reviewers agreed at journal 21 Apr, 2026 Reviewers invited by journal 20 Apr, 2026 Editor invited by journal 14 Apr, 2026 Editor assigned by journal 30 Mar, 2026 Submission checks completed at journal 30 Mar, 2026 First submitted to journal 29 Mar, 2026 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-9259481","acceptedTermsAndConditions":true,"allowDirectSubmit":false,"archivedVersions":[],"articleType":"Article","associatedPublications":[],"authors":[{"id":629861929,"identity":"71fce9a7-97cd-4981-95be-8bdc19a2112f","order_by":0,"name":"Xinning Liu","email":"","orcid":"","institution":"Dalian Neusoft University of Information","correspondingAuthor":false,"prefix":"","firstName":"Xinning","middleName":"","lastName":"Liu","suffix":""},{"id":629861930,"identity":"37616933-63bc-4a89-ba66-f5273bbf63b5","order_by":1,"name":"Li Han","email":"","orcid":"","institution":"Dalian Neusoft University of Information","correspondingAuthor":false,"prefix":"","firstName":"Li","middleName":"","lastName":"Han","suffix":""},{"id":629861931,"identity":"6edd6b70-b5c8-48d7-8088-4bc0d20047a9","order_by":2,"name":"Ling Kang","email":"","orcid":"","institution":"Dalian Neusoft University of Information","correspondingAuthor":false,"prefix":"","firstName":"Ling","middleName":"","lastName":"Kang","suffix":""},{"id":629861932,"identity":"2fa4c172-f831-441e-b472-755c4178bb34","order_by":3,"name":"Lei Zhao","email":"","orcid":"","institution":"Dalian Neusoft University of Information","correspondingAuthor":false,"prefix":"","firstName":"Lei","middleName":"","lastName":"Zhao","suffix":""},{"id":629861933,"identity":"4e1a1e8d-3cc7-4279-af55-206f5088185e","order_by":4,"name":"Quan Guo","email":"data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAAZAAAAAyAQMAAABI0h/eAAAABlBMVEX///8AAABVwtN+AAAACXBIWXMAAA7EAAAOxAGVKw4bAAAA6UlEQVRIiWNgGAWjYBADOQhlAyISiNNiDKHSDIjXkthAtBb5iOSDt3lq7qRvZ+8x/vAh4Q8DP3uOAcPPHbi1GN5IS7bmOfYsd2fPGTPJGQkGDJI9bwwYe8/g0TIjx0yah+1w7oYbOWbMvD8MGAxu5BgwM7YR0vLvcLrB/TfGn3mAttgT0iIvAdTC23Y4weAGj4E0SIuBBAEtBjzPki3n9h023HAmrQzoF2MeiTPPCg724rOlPfngjTffDssbHD+8GRhicnL87ckbH/zEZ8sBBgYJZAEeEHEAtwagLQ1oWkbBKBgFo2AUYAAAcetOgj8W14AAAAAASUVORK5CYII=","orcid":"","institution":"Dalian Neusoft University of Information","correspondingAuthor":true,"prefix":"","firstName":"Quan","middleName":"","lastName":"Guo","suffix":""},{"id":629861934,"identity":"36431878-9a0e-41e6-a5e4-74002a1ecfc9","order_by":5,"name":"Huadong Miao","email":"","orcid":"","institution":"Independent Researcher","correspondingAuthor":false,"prefix":"","firstName":"Huadong","middleName":"","lastName":"Miao","suffix":""}],"badges":[],"createdAt":"2026-03-29 14:38:18","currentVersionCode":1,"declarations":"","doi":"10.21203/rs.3.rs-9259481/v1","doiUrl":"https://doi.org/10.21203/rs.3.rs-9259481/v1","draftVersion":[],"editorialEvents":[],"editorialNote":"","failedWorkflow":false,"files":[{"id":108183334,"identity":"c8cd7cce-7409-4ab2-9f32-37b1919106ab","added_by":"auto","created_at":"2026-04-30 09:00:47","extension":"pdf","order_by":1,"title":"","display":"","copyAsset":false,"role":"manuscript-pdf","size":1378824,"visible":true,"origin":"","legend":"","description":"","filename":"Multiomicscancersubtypingviaattentionguidedvariationalrepresentationlearninganddeepclustering.pdf","url":"https://assets-eu.researchsquare.com/files/rs-9259481/v1_covered_27faa560-75f4-4a4b-bd0a-f15fdc820737.pdf"}],"financialInterests":"No competing interests reported.","formattedTitle":"Multi-omics cancer subtyping via attention-guided variational representation learning and deep clustering","fulltext":[],"fulltextSource":"","fullText":"","funders":[],"hasAdminPriorityOnWorkflow":false,"hasManuscriptDocX":false,"hasOptedInToPreprint":true,"hasPassedJournalQc":"","hasAnyPriority":false,"hideJournal":false,"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":"scientific-reports","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":false,"externalIdentity":"scirep","sideBox":"Learn more about [Scientific Reports](http://www.nature.com/srep/)","snPcode":"","submissionUrl":"","title":"Scientific Reports","twitterHandle":"","acdcEnabled":true,"dfaEnabled":true,"editorialSystem":"stoa","reportingPortfolio":"Scientific Reports","inReviewEnabled":true,"inReviewRevisionsEnabled":true},"keywords":"multi-omics integration, cancer subtyping, variational autoencoder, attention-based fusion, deep embedded clustering, survival stratification","lastPublishedDoi":"10.21203/rs.3.rs-9259481/v1","lastPublishedDoiUrl":"https://doi.org/10.21203/rs.3.rs-9259481/v1","license":{"name":"CC BY 4.0","url":"https://creativecommons.org/licenses/by/4.0/"},"manuscriptAbstract":"\u003cp\u003eTumor heterogeneity is a major challenge for accurate molecular subtyping and precision oncology, as complex variations across and within tumors can lead to distinct biological behaviors, treatment responses, and clinical outcomes. To address this issue, we propose MAVDC, a multi-omics integration framework that combines unsupervised feature selection, variational representation learning, attention-based fusion, contrastive learning, and adaptive deep embedded clustering. MAVDC first ranks features within each omics modality to suppress noise and retain informative signals, and then determines the optimal number of clusters using a composite scoring strategy. Compact variational autoencoders are employed to learn modality-specific latent representations, which are further integrated through a multi-head attention mechanism. A progressive training strategy jointly optimizes reconstruction, variational regularization, clustering refinement, and cross-omics consistency. Experiments on TCGA BRCA and LUSC cohorts integrating RNA expression, DNA methylation, and copy number variation data identified four molecular subtypes in each cancer. These subtypes exhibited clear geometric separation, distinct functional and immune characteristics, and significant survival differences, outperforming representative single-omics and multi-omics baseline methods. MAVDC provides an effective tool for discovering clinically meaningful cancer subtypes.\u003c/p\u003e","manuscriptTitle":"Multi-omics cancer subtyping via attention-guided variational representation learning and deep clustering","msid":"","msnumber":"","nonDraftVersions":[{"code":1,"date":"2026-04-29 07:10:55","doi":"10.21203/rs.3.rs-9259481/v1","editorialEvents":[{"type":"communityComments","content":0},{"type":"reviewerAgreed","content":"303345174014766522034044596934543817695","date":"2026-04-26T21:46:41+00:00","index":"hide","fulltext":""},{"type":"reviewerAgreed","content":"182448095844929728021685588822992479562","date":"2026-04-23T05:31:38+00:00","index":"hide","fulltext":""},{"type":"reviewerAgreed","content":"97180733154683610646130194224073132415","date":"2026-04-21T04:02:43+00:00","index":"hide","fulltext":""},{"type":"reviewersInvited","content":"","date":"2026-04-21T03:40:37+00:00","index":"","fulltext":""},{"type":"editorInvited","content":"","date":"2026-04-14T04:21:44+00:00","index":"","fulltext":""},{"type":"editorAssigned","content":"","date":"2026-03-30T07:05:47+00:00","index":"","fulltext":""},{"type":"checksComplete","content":"","date":"2026-03-30T07:05:42+00:00","index":"","fulltext":""},{"type":"submitted","content":"Scientific Reports","date":"2026-03-29T14:30:20+00:00","index":"","fulltext":""}],"status":"published","journal":{"display":true,"email":"[email protected]","identity":"scientific-reports","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":false,"externalIdentity":"scirep","sideBox":"Learn more about [Scientific Reports](http://www.nature.com/srep/)","snPcode":"","submissionUrl":"","title":"Scientific Reports","twitterHandle":"","acdcEnabled":true,"dfaEnabled":true,"editorialSystem":"stoa","reportingPortfolio":"Scientific Reports","inReviewEnabled":true,"inReviewRevisionsEnabled":true}}],"origin":"","ownerIdentity":"ac56ccdf-4b1b-49ce-bfaa-a27d4a6bd4b8","owner":[],"postedDate":"April 29th, 2026","published":true,"recentEditorialEvents":[],"rejectedJournal":[],"revision":"","amendment":"","status":"under-review","subjectAreas":[{"id":67036205,"name":"Biological sciences/Cancer"},{"id":67036206,"name":"Biological sciences/Computational biology and bioinformatics"}],"tags":[],"updatedAt":"2026-04-29T07:10:55+00:00","versionOfRecord":[],"versionCreatedAt":"2026-04-29 07:10:55","video":"","vorDoi":"","vorDoiUrl":"","workflowStages":[]},"version":"v1","identity":"rs-9259481","journalConfig":"researchsquare"},"__N_SSP":true},"page":"/article/[identity]/[[...version]]","query":{"redirect":"/article/rs-9259481","identity":"rs-9259481","version":["v1"]},"buildId":"XKTyCvWXoU3ODBz1xrDgd","isFallback":false,"isExperimentalCompile":false,"dynamicIds":[84888],"gssp":true,"scriptLoader":[]}