[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"branding":3,"analytics":7,"article-most-pubchem-bioassays-lack-basic-metadata-llms-could-fix":10,"sections":35},{"siteName":4,"siteTagline":5,"publisherName":4,"contactEmail":6},"The Revision","Tech news, decoded.","editor@therevision.news",{"gaMeasurementId":8,"adsenseClientId":9},"G-ZW2MV82GYR","ca-pub-8533917693782264",{"article":11},{"id":12,"slug":13,"title":14,"dek":15,"body_md":16,"tags_json":17,"published_at":18,"created_at":19,"updated_at":20,"status":21,"review_note":22,"review_notes":23,"image_url":22,"persona_id":22,"persona_name":22,"section":24,"tags":25,"sources":30,"feedback":34,"feedback_at":22,"cost_usd":34,"total_tokens":34},9746,"most-pubchem-bioassays-lack-basic-metadata-llms-could-fix","Most PubChem Bioassays Lack Basic Metadata LLMs Could Fix","A new study finds most of PubChem's two million bioassays lack standard metadata, and LLMs can recall missing labels with over 96 percent accuracy.","Most of PubChem's bioassays are missing basic metadata, and researchers think LLMs can fill in the gaps.\n\nA new study examined PubChem's roughly 2 million public bioassays, the screening experiments used to train AI models that predict how molecules behave. The authors found that 36 percent of those bioassays have no assay format tag, 89 percent have no BioAssay type, and more than 99.9 percent lack any mapping to the standard BioAssay Ontology terms for assay format or detection method. To see whether that gap could be automated away, they ran seven open-source and proprietary LLMs against the raw assay text and checked the models' guesses against existing labels from PubChem and ChEMBL. Recall came in at 0.96 or higher for both assay format and detection technology, and open-source models performed about as well as proprietary ones.\n\nThis is the unglamorous plumbing problem behind every AI-for-drug-discovery headline: foundation models for molecular property prediction are only as good as the metadata describing the experiments they are trained on, and that metadata has apparently been an afterthought for years. The more interesting finding is that a chunk of the disagreements between LLMs and the existing labels were not LLM mistakes at all - they traced back to inconsistencies in the human-curated data. In one case, showing an industrial curator the LLM's reasoning led that curator to revise their own earlier labels.\n\nThat is a genuinely useful job for LLMs: cheap auditors for institutional data nobody wants to re-check by hand. But the authors are upfront that per-class accuracy still varies and rare categories need human review before any of this feeds downstream models - a reminder that annotating millions of records is a much easier pitch than annotating all of them correctly.","[\"bioassay-annotation\",\"llm\",\"drug-discovery\",\"data-quality\"]","2026-10-02T04:00:00.000Z","2026-10-03T08:49:34.575Z","2026-10-03T08:49:40.584Z","published",null,[],"ai",[26,27,28,29],"bioassay-annotation","llm","drug-discovery","data-quality",[31],{"name":32,"url":33},"arXiv cs.AI","https:\u002F\u002Farxiv.org\u002Fabs\u002F2610.01616",0,{"sections":36},[37,40,44,48,53,57,61,66,71,76,81,86,91,96],{"name":38,"slug":24,"count":39,"latest_published_at":18},"AI",6041,{"name":41,"slug":42,"count":43,"latest_published_at":18},"Security","security",848,{"name":45,"slug":46,"count":47,"latest_published_at":18},"Policy","policy",439,{"name":49,"slug":50,"count":51,"latest_published_at":52},"Deals","deals",317,"2026-10-01T22:00:00.000Z",{"name":54,"slug":55,"count":56,"latest_published_at":18},"Hardware","hardware",199,{"name":58,"slug":59,"count":60,"latest_published_at":18},"Science","science",176,{"name":62,"slug":63,"count":64,"latest_published_at":65},"Consumer Tech","consumer-tech",155,"2026-10-01T19:54:10.000Z",{"name":67,"slug":68,"count":69,"latest_published_at":70},"Dev Tools","dev-tools",96,"2026-10-01T16:57:03.000Z",{"name":72,"slug":73,"count":74,"latest_published_at":75},"Software","software",93,"2026-09-30T21:41:11.000Z",{"name":77,"slug":78,"count":79,"latest_published_at":80},"Startups","startups",90,"2026-10-01T21:55:22.000Z",{"name":82,"slug":83,"count":84,"latest_published_at":85},"Gaming","gaming",53,"2026-10-02T02:50:39.000Z",{"name":87,"slug":88,"count":89,"latest_published_at":90},"General","general",50,"2026-09-30T21:37:54.000Z",{"name":92,"slug":93,"count":94,"latest_published_at":95},"Reviews","reviews",31,"2026-09-28T14:31:34.000Z",{"name":97,"slug":98,"count":99,"latest_published_at":100},"How-To","how-to",7,"2026-10-01T09:00:00.000Z"]