{"status":"ok","message-type":"work","message-version":"1.0.0","message":{"institution":[{"name":"bioRxiv"}],"indexed":{"date-parts":[[2026,10,23]],"date-time":"2026-10-23T21:19:20Z","timestamp":1792790360061,"version":"4.5.0"},"posted":{"date-parts":[[2026,9,21]]},"group-title":"Bioinformatics","reference-count":29,"publisher":"openRxiv","license":[{"start":{"date-parts":[[2026,9,21]],"date-time":"2026-09-21T00:00:00Z","timestamp":1789948800000},"content-version":"vor","delay-in-days":0,"URL":"http:\/\/creativecommons.org\/licenses\/by-nc-nd\/4.0\/"}],"funder":[{"award":["R01HG009937"],"award-info":[{"award-number":["R01HG009937"]}],"id":[{"id":"https:\/\/ror.org\/01cwqze88","id-type":"ROR","asserted-by":"publisher"}]},{"name":"Chan Zuckerberg Initiative","award":["2022-311195"],"award-info":[{"award-number":["2022-311195"]}],"id":[{"id":"https:\/\/ror.org\/02qenvm24","id-type":"ROR","asserted-by":"crossref"}]},{"name":"Chan Zuckerberg Initiative","award":["2024-342821"],"award-info":[{"award-number":["2024-342821"]}],"id":[{"id":"https:\/\/ror.org\/02qenvm24","id-type":"ROR","asserted-by":"crossref"}]}],"content-domain":{"domain":[],"crossmark-restriction":false},"short-container-title":[],"accepted":{"date-parts":[[2026,9,21]]},"abstract":"<jats:p>A cell-by-gene count matrix is the artifact of a single-cell RNA-seq experiment that is most often stored, shared, and reanalyzed. It is the output of a computation whose inputs are the sequenced molecules and a gene annotation, and while the molecules never change, the annotation is revised continually. Once the matrix has been produced, the evidence behind it can no longer be reinterpreted. Recovering that evidence means returning to raw reads or alignments that are large, costly to process, and frequently unavailable.<\/jats:p>\n                <jats:p>\n                  We ask whether a concise representation of the molecules themselves can be extracted once and reused indefinitely, to quantify under any future annotation, to query and discover features that no annotation yet describes, and to pool evidence across cells and samples. Our starting observation is that the procedures that turn alignments into counts, gene assignment and UMI collapse, never read most of what an alignment file contains. They consume\n                  <jats:italic>relations<\/jats:italic>\n                  among molecules such as shared genomic geometry, shared placements, barcode identity, and the equality or near-equality of UMIs. We show that these relations form a statistic that is sufficient for such consumers, and we design a compact, seekable, content-authenticated archive that stores them while deferring every annotation-dependent decision to analysis time. Archives compose into content-addressed collections that route cohort queries to the molecules that can answer them without copying molecules. We implement these ideas in a tool called\n                  <jats:monospace>gravlax<\/jats:monospace>\n                  .\n                <\/jats:p>\n                <jats:p>\n                  Across four human 10x 3\u2032 datasets,\n                  <jats:monospace>gravlax<\/jats:monospace>\n                  archives require 11\u201318 bits per read and are 9.0\u201312.7\u00d7 smaller than tag-preserving CRAM. Count matrices replayed from an archive deviate from direct\n                  <jats:monospace>STARsolo<\/jats:monospace>\n                  quantification by 0.24\u20130.75% of normalized UMI mass, whereas changing GENCODE v32 to v49 moves 2.12\u20134.64%, and quantification replay is 34\u2013 82\u00d7 faster than\n                  <jats:monospace>STARsolo<\/jats:monospace>\n                  at matched thread budgets. A federated index over eight archives occupies 2.96% of their size, answers a 96-query junction panel 2.59\u00d7 faster than the archives alone, and screens the cohort genome-wide for unannotated splice events that recur across donors in just 9 seconds.\n                <\/jats:p>\n                <jats:p>\n                  Because the molecules are retained, the archives also answer questions the matrix has discarded. An analysis of four peripheral-blood archives recovers a validated FYB1 immune-cell splicing switch, a cross-fitted fragment model appropriate for 3\u2032 chemistry reveals an eight-donor shift in NTRK2 terminal-isoform usage from astrocyte and neural-stem-cell populations to mature neurons, and pooling evidence across cells within the context of an expectation-maximization algorithm recovers 75\u201398% of withheld multi-gene molecule labels.\n                  <jats:monospace>Gravlax<\/jats:monospace>\n                  is open source, implemented in Rust, licensed under the BSD 3-clause license, and available at\n                  <jats:ext-link xmlns:xlink=\"http:\/\/www.w3.org\/1999\/xlink\" ext-link-type=\"uri\" xlink:href=\"https:\/\/github.com\/COMBINE-lab\/gravlax\">https:\/\/github.com\/COMBINE-lab\/gravlax<\/jats:ext-link>\n                  .\n                <\/jats:p>","DOI":"10.64898\/2026.09.18.752708","type":"posted-content","created":{"date-parts":[[2026,9,21]],"date-time":"2026-09-21T15:10:23Z","timestamp":1790003423000},"source":"Crossref","is-referenced-by-count":0,"title":["Gravlax: an annotation-independent molecular evidence archive for single-cell RNA-seq"],"prefix":"10.64898","author":[{"ORCID":"https:\/\/orcid.org\/0000-0001-8259-7434","authenticated-orcid":false,"given":"Rob","family":"Patro","sequence":"first","affiliation":[{"name":"Department of Computer Science, University of Maryland, College Park, MD 20742, United States"},{"name":"Center for Bioinformatics and Computational Biology, University of Maryland, College Park, MD 20742, United States"}],"role":[{"vocabulary":"crossref","role":"author"}]}],"member":"54368","reference":[{"key":"2026092415550990000_2026.09.18.752708v1.1","doi-asserted-by":"publisher","DOI":"10.1093\/nar\/gkac1071"},{"key":"2026092415550990000_2026.09.18.752708v1.2","doi-asserted-by":"crossref","first-page":"1506","DOI":"10.1038\/s41592-023-02003-w","article-title":"Recovery of missing single-cell RNA-sequencing data with optimized transcriptomic references","volume":"20","year":"2023","journal-title":"Nature Methods"},{"key":"2026092415550990000_2026.09.18.752708v1.3","doi-asserted-by":"crossref","first-page":"417","DOI":"10.1038\/nmeth.4197","article-title":"Salmon provides fast and biasaware quantification of transcript expression","volume":"14","year":"2017","journal-title":"Nature Methods"},{"key":"2026092415550990000_2026.09.18.752708v1.4","doi-asserted-by":"crossref","first-page":"316","DOI":"10.1038\/s41592-022-01408-3","article-title":"Alevin-fry unlocks rapid, accurate and memory-frugal quantification of single-cell RNA-seq data","volume":"19","year":"2022","journal-title":"Nature Methods"},{"key":"2026092415550990000_2026.09.18.752708v1.5","doi-asserted-by":"publisher","DOI":"10.1186\/s13059-021-02533-6"},{"key":"2026092415550990000_2026.09.18.752708v1.6","doi-asserted-by":"publisher","DOI":"10.1093\/bioinformatics\/btx547"},{"key":"2026092415550990000_2026.09.18.752708v1.7","doi-asserted-by":"publisher","DOI":"10.1093\/BIOINFORMATICS\/BTZ279"},{"key":"2026092415550990000_2026.09.18.752708v1.8","doi-asserted-by":"crossref","unstructured":"Dehghannasiri, R. et al. Reference-free discovery with barcoded single-cell sequencing. Nature Biotechnology (2026).","DOI":"10.1038\/s41587-026-03084-6"},{"key":"2026092415550990000_2026.09.18.752708v1.9","doi-asserted-by":"crossref","unstructured":"Le\u00f3n-Peri\u00f1\u00e1n, D. , Karaiskos, N. & Rajewsky, N. Ultrafast and reference-free sequence discovery in single-cell data. Nature (2026).","DOI":"10.1038\/s41586-026-10975-w"},{"key":"2026092415550990000_2026.09.18.752708v1.10","doi-asserted-by":"publisher","DOI":"10.1093\/nar\/gkw540"},{"key":"2026092415550990000_2026.09.18.752708v1.11","doi-asserted-by":"crossref","first-page":"14049","DOI":"10.1038\/ncomms14049","article-title":"Massively parallel digital transcriptional profiling of single cells","volume":"8","year":"2017","journal-title":"Nature Communications"},{"key":"2026092415550990000_2026.09.18.752708v1.12","doi-asserted-by":"publisher","DOI":"10.1101\/2021.05.05.442755"},{"key":"2026092415550990000_2026.09.18.752708v1.13","doi-asserted-by":"publisher","DOI":"10.1093\/bioinformatics\/bts635"},{"key":"2026092415550990000_2026.09.18.752708v1.14","unstructured":"Duda, J. Asymmetric numeral systems. https:\/\/arxiv.org\/abs\/0902.0271 (2009)."},{"key":"2026092415550990000_2026.09.18.752708v1.15","doi-asserted-by":"crossref","unstructured":"Collet, Y. & Kucherawy, M. Zstandard Compression and the Application\/zstd Media Type. https:\/\/www.rfc-editor.org\/rfc\/rfc8878 (2021).","DOI":"10.17487\/RFC8878"},{"key":"2026092415550990000_2026.09.18.752708v1.16","unstructured":"O\u2019Connor, J. , Aumasson, J.-P. , Neves, S. & Wilcox-O\u2019Hearn, Z. BLAKE3: one function, fast everywhere. (2020)."},{"key":"2026092415550990000_2026.09.18.752708v1.17","doi-asserted-by":"publisher","DOI":"10.1101\/gr.209601.116"},{"key":"2026092415550990000_2026.09.18.752708v1.18","doi-asserted-by":"publisher","DOI":"10.1093\/nar\/gkz918"},{"key":"2026092415550990000_2026.09.18.752708v1.19","doi-asserted-by":"publisher","DOI":"10.1093\/bioinformatics\/btac010"},{"key":"2026092415550990000_2026.09.18.752708v1.20","doi-asserted-by":"crossref","first-page":"5233","DOI":"10.1038\/s41598-019-41695-z","article-title":"From Louvain to Leiden: guaranteeing well-connected communities","volume":"9","year":"2019","journal-title":"Scientific Reports"},{"key":"2026092415550990000_2026.09.18.752708v1.21","doi-asserted-by":"publisher","DOI":"10.1186\/s13059-017-1382-0"},{"key":"2026092415550990000_2026.09.18.752708v1.22","doi-asserted-by":"publisher","DOI":"10.3724\/SP.J.1087.2009.02191"},{"key":"2026092415550990000_2026.09.18.752708v1.23","doi-asserted-by":"publisher","DOI":"10.1038\/s41586-018-0414-6"},{"key":"2026092415550990000_2026.09.18.752708v1.24","doi-asserted-by":"crossref","unstructured":"Puvogel, S. et al. Single-Nucleus RNA-Seq Characterizes the Cell Types Along the Neuronal Lineage in the Adult Human Subependymal Zone and Reveals Reduced Oligodendrocyte Progenitor Abundance with Age. eNeuro 11, ENEURO.246\u201323.2024 (2024).","DOI":"10.1523\/ENEURO.0246-23.2024"},{"key":"2026092415550990000_2026.09.18.752708v1.25","doi-asserted-by":"crossref","first-page":"1116392","DOI":"10.3389\/fimmu.2023.1116392","article-title":"Mapping the splicing landscape of the human immune system","volume":"14","year":"2023","journal-title":"Frontiers in Immunology"},{"key":"2026092415550990000_2026.09.18.752708v1.26","doi-asserted-by":"crossref","first-page":"307","DOI":"10.1038\/s41592-022-01400-x","article-title":"The SpliZ generalizes percent spliced in to reveal regulated splicing at single-cell resolution","volume":"19","year":"2022","journal-title":"Nature Methods"},{"key":"2026092415550990000_2026.09.18.752708v1.27","doi-asserted-by":"publisher","DOI":"10.1038\/s41586-022-04558-8"},{"key":"2026092415550990000_2026.09.18.752708v1.28","doi-asserted-by":"publisher","DOI":"10.7554\/eLife.44667"},{"key":"2026092415550990000_2026.09.18.752708v1.29","unstructured":"10x Genomics. 10k Human Brain Nuclei, Chromium GEM-X Epi Multiome. https:\/\/www.10xgenomics.com\/datasets\/multiome-gemx-10k-human-brain (2026)."}],"container-title":[],"original-title":[],"link":[{"URL":"https:\/\/syndication.highwire.org\/content\/doi\/10.64898\/2026.09.18.752708","content-type":"unspecified","content-version":"vor","intended-application":"similarity-checking"}],"deposited":{"date-parts":[[2026,9,24]],"date-time":"2026-09-24T22:55:18Z","timestamp":1790290518000},"score":1,"resource":{"primary":{"URL":"http:\/\/biorxiv.org\/lookup\/doi\/10.64898\/2026.09.18.752708"}},"subtitle":[],"short-title":[],"issued":{"date-parts":[[2026,9,21]]},"references-count":29,"URL":"https:\/\/doi.org\/10.64898\/2026.09.18.752708","relation":{},"subject":[],"published":{"date-parts":[[2026,9,21]]},"subtype":"preprint"}}