Filter-Guided Lasso Regression in Gene Expression Data: A Systematic Investigation of the Stability–Accuracy Tradeoff Across Dimensionality Regimes.

Filter-Guided Lasso Regression in Gene Expression Data: A Systematic Investigation of the Stability–Accuracy Tradeoff Across Dimensionality 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 Filter-Guided Lasso Regression in Gene Expression Data: A Systematic Investigation of the Stability–Accuracy Tradeoff Across Dimensionality Regimes. Adel Aloraini This is a preprint; it has not been peer reviewed by a journal. https://doi.org/ 10.21203/rs.3.rs-9271860/v1 This work is licensed under a CC BY 4.0 License Status: Under Review Version 1 posted 10 You are reading this latest preprint version Abstract Background: Lasso regression is widely used for feature selection in high-dimensional gene expression data, but its selections are unstable across data perturbations, undermining reproducibility. Univariate filter-based pre-screening is commonly applied before regularised regression, yet the interaction between filter choice, rank aggregation strategy, pre-screening stringency, and the resulting stability of lasso feature selection has not been systematically characterised. Results: We evaluated 79 filter–lasso configurations—combining six univariate filters (Pearson, Spearman, Kendall correlation, mutual information, F-statistic, distance correlation), five rank aggregation strategies, and six pre-screening thresholds— against an unfiltered lasso baseline across six gene expression datasets spanning p/n ratios from 0.46 to 1.35. Filter-guided pre-screening produced 2–2.5 times higher bootstrap selection stability and 25–37% fewer selected predictors, at a modest cost of 1–2% in cross-validated R 2 in high-dimensional datasets. In the low-dimensional regime, pre-screening additionally improved predictive accuracy by up to 7 percentage points. We discovered that Pearson correlation, Gaussian mutual information, and the F-statistic produce identical gene rankings across all six datasets, revealing that only three functionally distinct filter families exist among the six evaluated.The choice of pre-screening threshold mattered more than the choice of filter oraggregation method. Conclusions: Filter-guided pre-screening for lasso regression is best understood as a mechanism for improving reproducibility and parsimony rather than prediction. It is most beneficial when model interpretability is prioritised, with strongest advantages in lower-dimensional settings. Practitioners need only three filter types—linear,rank-based, and nonlinear—to capture all available pre-screening information. Lasso regression Feature selection stability Univariate filters Rank aggregation Gene expression High-dimensional data Reproducibility Full Text Additional Declarations No competing interests reported. Supplementary Files Codes.zip Cite Share Download PDF Status: Under Review Version 1 posted Reviewers agreed at journal 28 Apr, 2026 Reviewers agreed at journal 25 Apr, 2026 Reviewers agreed at journal 24 Apr, 2026 Reviewers agreed at journal 23 Apr, 2026 Reviewers agreed at journal 23 Apr, 2026 Reviewers invited by journal 23 Apr, 2026 Editor assigned by journal 01 Apr, 2026 Editor invited by journal 01 Apr, 2026 Submission checks completed at journal 31 Mar, 2026 First submitted to journal 31 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-9271860","acceptedTermsAndConditions":true,"allowDirectSubmit":false,"archivedVersions":[],"articleType":"Research Article","associatedPublications":[],"authors":[{"id":630545254,"identity":"e36b6277-34a3-4785-b3ec-d18b72e669f3","order_by":0,"name":"Adel Aloraini","email":"data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAAZAAAAAyAQMAAABI0h/eAAAABlBMVEX///8AAABVwtN+AAAACXBIWXMAAA7EAAAOxAGVKw4bAAAA00lEQVRIiWNgGAWjYNACAwYefjCDjRQtkg2kaQHpOkCsFv4G5mMffhTckTE+nmPA8KHsMIN8+wH8WiQOsCXP7DF4xmN25o0B44xzhxkMziQQsOYAjzEDj8FhHrMbOQbMvG1ALQwEtMgDtTD+AWoxngHU8heoRb7/AX4tBkAtzCBbDCSAWhiBWhhuELDF8DBbMrMMUIvEmWcFB3vOpfMY3CBgi9zx5sOMb/4ctudvT9744EeZtZx8PwFbGJjhrASGA0CSh4B6FEDI8FEwCkbBKBixAACkFD3m+8XdpgAAAABJRU5ErkJggg==","orcid":"","institution":"Qassim University","correspondingAuthor":true,"prefix":"","firstName":"Adel","middleName":"","lastName":"Aloraini","suffix":""}],"badges":[],"createdAt":"2026-03-30 20:53:14","currentVersionCode":1,"declarations":"","doi":"10.21203/rs.3.rs-9271860/v1","doiUrl":"https://doi.org/10.21203/rs.3.rs-9271860/v1","draftVersion":[],"editorialEvents":[],"editorialNote":"","failedWorkflow":false,"files":[{"id":108006960,"identity":"1aa67795-4d77-4aa9-af3a-c984fd442827","added_by":"auto","created_at":"2026-04-28 12:58:03","extension":"pdf","order_by":1,"title":"","display":"","copyAsset":false,"role":"manuscript-pdf","size":361206,"visible":true,"origin":"","legend":"","description":"","filename":"BMCBioinformatics.pdf","url":"https://assets-eu.researchsquare.com/files/rs-9271860/v1_covered_e160088c-1e69-477e-8e1a-5a1b5d863244.pdf"},{"id":107995365,"identity":"4e695fee-03d5-4647-b54d-adf287c3a4dd","added_by":"auto","created_at":"2026-04-28 11:04:26","extension":"zip","order_by":0,"title":"","display":"","copyAsset":false,"role":"supplement","size":44556,"visible":true,"origin":"","legend":"","description":"","filename":"Codes.zip","url":"https://assets-eu.researchsquare.com/files/rs-9271860/v1/ddca1077fceac256b31dc1cd.zip"}],"financialInterests":"No competing interests reported.","formattedTitle":"Filter-Guided Lasso Regression in Gene Expression Data: A Systematic Investigation of the Stability–Accuracy Tradeoff Across Dimensionality Regimes.","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":"bmc-bioinformatics","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":false,"externalIdentity":"binf","sideBox":"Learn more about [BMC Bioinformatics](http://bmcbioinformatics.biomedcentral.com/)","snPcode":"","submissionUrl":"https://www.editorialmanager.com/binf","title":"BMC Bioinformatics","twitterHandle":"@BMC_Bioinformatics","acdcEnabled":true,"dfaEnabled":false,"editorialSystem":"em","reportingPortfolio":"BMC Series","inReviewEnabled":true,"inReviewRevisionsEnabled":true},"keywords":"Lasso regression, Feature selection stability, Univariate filters, Rank aggregation, Gene expression, High-dimensional data, Reproducibility","lastPublishedDoi":"10.21203/rs.3.rs-9271860/v1","lastPublishedDoiUrl":"https://doi.org/10.21203/rs.3.rs-9271860/v1","license":{"name":"CC BY 4.0","url":"https://creativecommons.org/licenses/by/4.0/"},"manuscriptAbstract":"\u003cp\u003eBackground: Lasso regression is widely used for feature selection in high-dimensional\u0026nbsp;gene expression data, but its selections are unstable across data perturbations, undermining reproducibility. Univariate filter-based pre-screening is commonly applied\u0026nbsp;before regularised regression, yet the interaction between filter choice, rank aggregation strategy, pre-screening stringency, and the resulting stability of lasso feature\u0026nbsp;selection has not been systematically characterised.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eResults: We evaluated 79 filter–lasso configurations—combining six univariate filters\u0026nbsp;(Pearson, Spearman, Kendall correlation, mutual information, F-statistic, distance\u0026nbsp;correlation), five rank aggregation strategies, and six pre-screening thresholds— \u0026nbsp;against an unfiltered lasso baseline across six gene expression datasets spanning p/n\u0026nbsp;ratios from 0.46 to 1.35. Filter-guided pre-screening produced 2–2.5 times higher \u0026nbsp;bootstrap selection stability and 25–37% fewer selected predictors, at a modest cost of 1–2% in cross-validated R\u003csup\u003e2\u003c/sup\u003e in high-dimensional datasets. In the low-dimensional\u0026nbsp;regime, pre-screening additionally improved predictive accuracy by up to 7 percentage points. We discovered that Pearson correlation, Gaussian mutual information, and\u0026nbsp;the F-statistic produce identical gene rankings across all six datasets, revealing that only three functionally distinct filter families exist among the six evaluated.The choice of pre-screening threshold mattered more than the choice of filter oraggregation method. \u0026nbsp;\u003c/p\u003e\n\u003cp\u003eConclusions: Filter-guided pre-screening for lasso regression is best understood as a\u0026nbsp;mechanism for improving reproducibility and parsimony rather than prediction. It is\u0026nbsp;most beneficial when model interpretability is prioritised, with strongest advantages \u0026nbsp;in lower-dimensional settings. Practitioners need only three filter types—linear,rank-based, and nonlinear—to capture all available pre-screening information. \u0026nbsp;\u003c/p\u003e","manuscriptTitle":"Filter-Guided Lasso Regression in Gene Expression Data: A Systematic Investigation of the Stability–Accuracy Tradeoff Across Dimensionality Regimes.","msid":"","msnumber":"","nonDraftVersions":[{"code":1,"date":"2026-04-28 11:04:22","doi":"10.21203/rs.3.rs-9271860/v1","editorialEvents":[{"type":"communityComments","content":0},{"type":"reviewerAgreed","content":"140717718115190785422877135843877135332","date":"2026-04-28T19:20:58+00:00","index":"hide","fulltext":""},{"type":"reviewerAgreed","content":"229811903023268535624550010326449231576","date":"2026-04-25T08:36:04+00:00","index":"hide","fulltext":""},{"type":"reviewerAgreed","content":"258582603223603956118734078068342945616","date":"2026-04-24T11:04:40+00:00","index":"hide","fulltext":""},{"type":"reviewerAgreed","content":"219300427416397488616434210112400950893","date":"2026-04-23T13:36:58+00:00","index":"hide","fulltext":""},{"type":"reviewerAgreed","content":"261872873084191358998642731614806185952","date":"2026-04-23T11:14:28+00:00","index":"hide","fulltext":""},{"type":"reviewersInvited","content":"","date":"2026-04-23T06:40:04+00:00","index":"","fulltext":""},{"type":"editorAssigned","content":"","date":"2026-04-01T07:53:44+00:00","index":"","fulltext":""},{"type":"editorInvited","content":"","date":"2026-04-01T07:37:25+00:00","index":"","fulltext":""},{"type":"checksComplete","content":"","date":"2026-03-31T18:11:01+00:00","index":"","fulltext":""},{"type":"submitted","content":"BMC Bioinformatics","date":"2026-03-31T12:17:33+00:00","index":"","fulltext":""}],"status":"published","journal":{"display":true,"email":"[email protected]","identity":"bmc-bioinformatics","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":false,"externalIdentity":"binf","sideBox":"Learn more about [BMC Bioinformatics](http://bmcbioinformatics.biomedcentral.com/)","snPcode":"","submissionUrl":"https://www.editorialmanager.com/binf","title":"BMC Bioinformatics","twitterHandle":"@BMC_Bioinformatics","acdcEnabled":true,"dfaEnabled":false,"editorialSystem":"em","reportingPortfolio":"BMC Series","inReviewEnabled":true,"inReviewRevisionsEnabled":true}}],"origin":"","ownerIdentity":"473cfcf9-ee76-4651-9fc7-c95bd2324270","owner":[],"postedDate":"April 28th, 2026","published":true,"recentEditorialEvents":[],"rejectedJournal":[],"revision":"","amendment":"","status":"under-review","subjectAreas":[],"tags":[],"updatedAt":"2026-04-28T11:04:22+00:00","versionOfRecord":[],"versionCreatedAt":"2026-04-28 11:04:22","video":"","vorDoi":"","vorDoiUrl":"","workflowStages":[]},"version":"v1","identity":"rs-9271860","journalConfig":"researchsquare"},"__N_SSP":true},"page":"/article/[identity]/[[...version]]","query":{"redirect":"/article/rs-9271860","identity":"rs-9271860","version":["v1"]},"buildId":"XKTyCvWXoU3ODBz1xrDgd","isFallback":false,"isExperimentalCompile":false,"dynamicIds":[84888],"gssp":true,"scriptLoader":[]}