Long-context DNA models are limited by token-mixing cost and by how compression allocates representational budget across the genome. Existing approaches operate close to base-pair resolution, apply fixed downsampling, or learn content-dependent chunks without an explicit genomic budget, making long-context pretraining expensive and difficult to control. We introduce GeneZip, a region-aware DNA compression framework that combines H-Net-style dynamic routing with a Region-Aware Ratio (RAR) objective and bounded routing. GeneZip uses static gene-structure annotations during compression training to specify region-wise base-pairs-per-token (BPT) targets; at inference time, it compresses raw unseen DNA without annotations. GeneZip provides three main benefits. First, it is effective: GeneZip variants achieve the best validation PPL among encoder-based compressors, with GeneZip-70M operating at 137.6 BPT, and across four reproducible DNALongBench tasks–contact map prediction, eQTL prediction, enhancer-target gene prediction, and transcription-initiation signal prediction–GeneZip obtains the best average rank among compared sequence models. Second, it is redundancy-aware: a post-hoc RepeatMasker\/TRF analysis shows that, without repeat supervision, GeneZip assigns higher local BPT to TE-derived interspersed repeats and tandem repeats, two major classes of repetitive DNA sequence redundancy. Third, it is efficient: by reducing the effective token-mixing length, GeneZip enables longer-context and larger-capacity pretraining, including 128K-context and 636M-parameter variants on a single A100 80GB GPU, and fine-tunes the eQTL task 50.4x faster than JanusDNA (50 vs. 2520 minutes). These results establish GeneZip as an effective, redundancy-aware, and efficient compression interface for long-context DNA modeling.<\/p>\n","protected":false},"excerpt":{"rendered":"
Long-context DNA models are limited by token-mixing cost and by how compression allocates representational budget across the genome. Existing approaches operate close to base-pair resolution, apply fixed downsampling, or learn content-dependent chunks without an explicit genomic budget, making long-context pretraining expensive and difficult to control. We introduce GeneZip, a region-aware DNA compression framework that combines […]<\/p>\n","protected":false},"featured_media":0,"template":"","meta":{"msr-url-field":"","msr-podcast-episode":"","msrModifiedDate":"","msrModifiedDateEnabled":false,"ep_exclude_from_search":false,"_classifai_error":"","msr-author-ordering":[{"type":"text","value":"Jianan Zhao","user_id":0},{"type":"text","value":"Xixian Liu","user_id":0},{"type":"text","value":"Zhihao Zhan","user_id":0},{"type":"text","value":"Xinyu Yuan","user_id":0},{"type":"text","value":"Hongyu Guo","user_id":0},{"type":"user_nicename","value":"Jian Tang","user_id":"32270"}],"msr_publishername":"","msr_publisher_other":"","msr_booktitle":"","msr_chapter":"","msr_edition":"","msr_editors":"","msr_how_published":"","msr_isbn":"","msr_issue":"","msr_journal":"","msr_number":"","msr_organization":"","msr_pages_string":"","msr_page_range_start":"","msr_page_range_end":"","msr_series":"","msr_volume":"","msr_copyright":"","msr_conference_name":"NeurIPS 2026","msr_doi":"10.48550\/arXiv.2602.17739","msr_arxiv_id":"2602.17739","msr_mag_id":"","msr_other_authors":"","msr_other_contributors":"","msr_speaker":"","msr_award":"","msr_affiliation":"","msr_institution":"","msr_host":"","msr_version":"","msr_duration":"","msr_release_tracker_id":"","msr_highlight_type":"","msr_date_display_format":"","msr_main_download_label":"","msr_external_link_label":"","msr_doi_label":"","msr_published_date":"2026-02-19","msr_startdate":"","msr_presentation_date":"","msr_highlight_text":"","msr_notes":"","msr_longbiography":"","msr_publicationurl":"","msr_external_url":"","msr_secondary_video_url":"","msr_conference_url":"","msr_journal_url":"","msr_year":2026,"msr_month":2,"msr_day":19,"msr_microsoftintellectualproperty":false,"msr_pub_id":"510aef266a92ca2a05d8d6e880d27ad35d6a9b0d","msr_publication_uploader":[{"type":"url","title":"https:\/\/arxiv.org\/abs\/2602.17739","label_id":243109,"id":false,"viewUrl":false},{"type":"url","title":"https:\/\/dblp.org\/rec\/journals\/corr\/abs-2602-17739.html","label_id":243109,"id":false,"viewUrl":false}],"msr_related_uploader":[],"msr_original_fields_of_study":[],"msr_s2_paper_id":"510aef266a92ca2a05d8d6e880d27ad35d6a9b0d","msr_s2_pdf_url":"","msr_citation_count_updated":"","msr_citation_count":0,"msr_influential_citations":0,"msr_reference_count":54,"msr_s2_open_access":false,"msr_s2_author_ids":[],"msr_pub_ids":[{"provider":"s2","id":"510aef266a92ca2a05d8d6e880d27ad35d6a9b0d"},{"provider":"doi","id":"10.48550\/arXiv.2602.17739"},{"provider":"arxiv","id":"2602.17739"},{"provider":"dblp","id":"journals\/corr\/abs-2602-17739"},{"provider":"corpusid","id":"285872688"}],"msr_hide_image_in_river":0,"footnotes":""},"msr-research-highlight":[],"research-area":[13553],"msr-publication-type":[193716],"msr-publisher":[],"msr-publication-cta":[],"msr-focus-area":[],"msr-locale":[268875],"msr-post-option":[],"msr-field-of-study":[250972,246691],"msr-conference":[259048],"msr-journal":[268188],"msr-impact-theme":[],"msr-pillar":[],"class_list":["post-1191494","msr-research-item","type-msr-research-item","status-publish","hentry","msr-research-area-medical-health-genomics","msr-locale-en_us","msr-field-of-study-biology","msr-field-of-study-computer-science"],"msr_publishername":"","msr_edition":"","msr_affiliation":"","msr_published_date":"2026-02-19","msr_host":"","msr_duration":"","msr_version":"","msr_speaker":"","msr_other_contributors":"","msr_booktitle":"","msr_pages_string":"","msr_chapter":"","msr_isbn":"","msr_journal":"","msr_volume":"","msr_number":"","msr_editors":"","msr_series":"","msr_issue":"","msr_organization":"","msr_how_published":"","msr_notes":"","msr_highlight_text":"","msr_release_tracker_id":"","msr_original_fields_of_study":"","msr_download_urls":"","msr_external_url":"","msr_secondary_video_url":"","msr_longbiography":"","msr_microsoftintellectualproperty":0,"msr_main_download":"","msr_publicationurl":"","msr_doi":"10.48550\/arXiv.2602.17739","msr_publication_uploader":[{"type":"url","title":"https:\/\/arxiv.org\/abs\/2602.17739","label_id":243109,"id":false,"viewUrl":false},{"type":"url","title":"https:\/\/dblp.org\/rec\/journals\/corr\/abs-2602-17739.html","label_id":243109,"id":false,"viewUrl":false}],"msr_related_uploader":[],"msr_citation_count":0,"msr_citation_count_updated":"","msr_s2_paper_id":"510aef266a92ca2a05d8d6e880d27ad35d6a9b0d","msr_influential_citations":0,"msr_reference_count":54,"msr_arxiv_id":"2602.17739","msr_s2_author_ids":[],"msr_s2_open_access":false,"msr_s2_pdf_url":null,"msr_attachments":[],"msr-author-ordering":[{"type":"text","value":"Jianan Zhao","user_id":0,"rest_url":false},{"type":"text","value":"Xixian Liu","user_id":0,"rest_url":false},{"type":"text","value":"Zhihao Zhan","user_id":0,"rest_url":false},{"type":"text","value":"Xinyu Yuan","user_id":0,"rest_url":false},{"type":"text","value":"Hongyu Guo","user_id":0,"rest_url":false},{"type":"user_nicename","value":"Jian Tang","user_id":32270,"rest_url":"https:\/\/cm-edgetun.pages.dev\/en-us\/research\/wp-json\/microsoft-research\/v1\/researchers?person=Jian Tang"}],"msr_impact_theme":[],"msr_research_lab":[],"msr_event":[],"msr_group":[],"msr_project":[],"publication":[],"video":[],"msr-tool":[],"msr_publication_type":"inproceedings","related_content":[],"_links":{"self":[{"href":"https:\/\/cm-edgetun.pages.dev\/en-us\/research\/wp-json\/wp\/v2\/msr-research-item\/1191494","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/cm-edgetun.pages.dev\/en-us\/research\/wp-json\/wp\/v2\/msr-research-item"}],"about":[{"href":"https:\/\/cm-edgetun.pages.dev\/en-us\/research\/wp-json\/wp\/v2\/types\/msr-research-item"}],"version-history":[{"count":4,"href":"https:\/\/cm-edgetun.pages.dev\/en-us\/research\/wp-json\/wp\/v2\/msr-research-item\/1191494\/revisions"}],"predecessor-version":[{"id":1192231,"href":"https:\/\/cm-edgetun.pages.dev\/en-us\/research\/wp-json\/wp\/v2\/msr-research-item\/1191494\/revisions\/1192231"}],"wp:attachment":[{"href":"https:\/\/cm-edgetun.pages.dev\/en-us\/research\/wp-json\/wp\/v2\/media?parent=1191494"}],"wp:term":[{"taxonomy":"msr-research-highlight","embeddable":true,"href":"https:\/\/cm-edgetun.pages.dev\/en-us\/research\/wp-json\/wp\/v2\/msr-research-highlight?post=1191494"},{"taxonomy":"msr-research-area","embeddable":true,"href":"https:\/\/cm-edgetun.pages.dev\/en-us\/research\/wp-json\/wp\/v2\/research-area?post=1191494"},{"taxonomy":"msr-publication-type","embeddable":true,"href":"https:\/\/cm-edgetun.pages.dev\/en-us\/research\/wp-json\/wp\/v2\/msr-publication-type?post=1191494"},{"taxonomy":"msr-publisher","embeddable":true,"href":"https:\/\/cm-edgetun.pages.dev\/en-us\/research\/wp-json\/wp\/v2\/msr-publisher?post=1191494"},{"taxonomy":"msr-publication-cta","embeddable":true,"href":"https:\/\/cm-edgetun.pages.dev\/en-us\/research\/wp-json\/wp\/v2\/msr-publication-cta?post=1191494"},{"taxonomy":"msr-focus-area","embeddable":true,"href":"https:\/\/cm-edgetun.pages.dev\/en-us\/research\/wp-json\/wp\/v2\/msr-focus-area?post=1191494"},{"taxonomy":"msr-locale","embeddable":true,"href":"https:\/\/cm-edgetun.pages.dev\/en-us\/research\/wp-json\/wp\/v2\/msr-locale?post=1191494"},{"taxonomy":"msr-post-option","embeddable":true,"href":"https:\/\/cm-edgetun.pages.dev\/en-us\/research\/wp-json\/wp\/v2\/msr-post-option?post=1191494"},{"taxonomy":"msr-field-of-study","embeddable":true,"href":"https:\/\/cm-edgetun.pages.dev\/en-us\/research\/wp-json\/wp\/v2\/msr-field-of-study?post=1191494"},{"taxonomy":"msr-conference","embeddable":true,"href":"https:\/\/cm-edgetun.pages.dev\/en-us\/research\/wp-json\/wp\/v2\/msr-conference?post=1191494"},{"taxonomy":"msr-journal","embeddable":true,"href":"https:\/\/cm-edgetun.pages.dev\/en-us\/research\/wp-json\/wp\/v2\/msr-journal?post=1191494"},{"taxonomy":"msr-impact-theme","embeddable":true,"href":"https:\/\/cm-edgetun.pages.dev\/en-us\/research\/wp-json\/wp\/v2\/msr-impact-theme?post=1191494"},{"taxonomy":"msr-pillar","embeddable":true,"href":"https:\/\/cm-edgetun.pages.dev\/en-us\/research\/wp-json\/wp\/v2\/msr-pillar?post=1191494"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}