{"status":"ok","message-type":"work","message-version":"1.0.0","message":{"institution":[{"name":"bioRxiv"}],"indexed":{"date-parts":[[2026,9,30]],"date-time":"2026-09-30T16:56:56Z","timestamp":1790787416786,"version":"4.1.0"},"posted":{"date-parts":[[2026,9,27]]},"group-title":"Bioinformatics","reference-count":20,"publisher":"openRxiv","license":[{"start":{"date-parts":[[2026,9,27]],"date-time":"2026-09-27T00:00:00Z","timestamp":1790467200000},"content-version":"vor","delay-in-days":0,"URL":"http:\/\/creativecommons.org\/licenses\/by\/4.0\/"}],"content-domain":{"domain":[],"crossmark-restriction":false},"short-container-title":[],"accepted":{"date-parts":[[2026,9,27]]},"abstract":"<jats:title>Abstract<\/jats:title>\n                <jats:p>Advances in protein structure prediction have enabled all-atom protein-ligand co-folding models that predict bound conformations directly from sequence and small-molecule structure. However, these models often fail to generalize to novel binding sites or alternative protein conformational states, limiting their utility for chemical biology and drug discovery. Here we show this limitation reflects training data bias rather than architectural constraints and can be overcome through targeted finetuning. Using ten previously unseen X-ray structures of Werner (WRN) helicase from a drug discovery program, we finetune Boltz-1 to learn both an allosteric binding site and a large conformational change locking the enzyme in an inactive state, while preserving accuracy on the ATP-bound state. The finetuned model generalizes to different chemical series and transfers the conformational logic across RecQ-family helicases in a binding-site sequence-dependent manner. This approach provides a blueprint for adapting foundation models as new structural and mechanistic data emerge, enabling co-folding networks to capture ligand-induced conformational switches and binding poses absent from their training data but central to biological regulation and therapeutic intervention.<\/jats:p>","DOI":"10.64898\/2026.09.21.752570","type":"posted-content","created":{"date-parts":[[2026,9,28]],"date-time":"2026-09-28T04:35:13Z","timestamp":1790570113000},"source":"Crossref","is-referenced-by-count":0,"title":["Targeted finetuning enables co-folding models to learn ligand-induced protein conformational states"],"prefix":"10.64898","author":[{"ORCID":"https:\/\/orcid.org\/0000-0003-4344-0383","authenticated-orcid":false,"given":"Rohan","family":"Gorantla","sequence":"first","affiliation":[{"name":"Novartis Biomedical Research, Discovery Sciences, Basel 4056, Switzerland"}],"role":[{"vocabulary":"crossref","role":"author"}]},{"ORCID":"https:\/\/orcid.org\/0000-0002-5601-1724","authenticated-orcid":false,"given":"Christian","family":"Schleberger","sequence":"additional","affiliation":[{"name":"Novartis Biomedical Research, Discovery Sciences, Basel 4056, Switzerland"}],"role":[{"vocabulary":"crossref","role":"author"}]},{"ORCID":"https:\/\/orcid.org\/0000-0001-8331-4344","authenticated-orcid":false,"given":"Fabian","family":"Sesterhenn","sequence":"additional","affiliation":[{"name":"Novartis Biomedical Research, Discovery Sciences, Basel 4056, Switzerland"}],"role":[{"vocabulary":"crossref","role":"author"}]}],"member":"54368","reference":[{"key":"2026093008551220000_2026.09.21.752570v1.1","doi-asserted-by":"publisher","DOI":"10.1038\/S41586-024-07487-W"},{"key":"2026093008551220000_2026.09.21.752570v1.2","doi-asserted-by":"publisher","DOI":"10.1038\/s41592-024-02272-z"},{"key":"2026093008551220000_2026.09.21.752570v1.3","doi-asserted-by":"publisher","DOI":"10.1101\/2024.11.19.624167"},{"key":"2026093008551220000_2026.09.21.752570v1.4","doi-asserted-by":"publisher","DOI":"10.1101\/2024.10.10.615955"},{"key":"2026093008551220000_2026.09.21.752570v1.5","doi-asserted-by":"publisher","DOI":"10.64898\/2026.02.05.703733"},{"key":"2026093008551220000_2026.09.21.752570v1.6","doi-asserted-by":"crossref","unstructured":"Nittinger, E. , Yoluk, \u00d6. , Tibo, A. , Olanders, G. & Tyrchan, C . Co-folding, the future of docking \u2013 prediction of allosteric and orthosteric ligands. Artificial Intelligence in the Life Sciences 8, 100136 (2025).","DOI":"10.1016\/j.ailsci.2025.100136"},{"key":"2026093008551220000_2026.09.21.752570v1.7","doi-asserted-by":"publisher","DOI":"10.1101\/2025.02.03.636309"},{"key":"2026093008551220000_2026.09.21.752570v1.8","doi-asserted-by":"publisher","DOI":"10.1038\/s41467-025-63947-5"},{"key":"2026093008551220000_2026.09.21.752570v1.9","doi-asserted-by":"publisher","DOI":"10.1038\/s41586-019-1102-x"},{"key":"2026093008551220000_2026.09.21.752570v1.10","doi-asserted-by":"publisher","DOI":"10.26508\/lsa.202000795"},{"key":"2026093008551220000_2026.09.21.752570v1.11","doi-asserted-by":"publisher","DOI":"10.1038\/s41586-024-07350-y"},{"key":"2026093008551220000_2026.09.21.752570v1.12","doi-asserted-by":"publisher","DOI":"10.1101\/2025.06.14.659707"},{"key":"2026093008551220000_2026.09.21.752570v1.13","unstructured":"Hu, E. J. , et al. LoRA: Low-Rank Adaptation of Large Language Models. (2021)."},{"key":"2026093008551220000_2026.09.21.752570v1.14","doi-asserted-by":"crossref","first-page":"387","DOI":"10.1038\/s41592-025-02973-z","article-title":"A fully automated benchmarking suite to compare macromolecular complexes","volume":"23","year":"2026","journal-title":"Nat. Methods"},{"key":"2026093008551220000_2026.09.21.752570v1.15","doi-asserted-by":"crossref","unstructured":"Palte, R. L. et al. WRN structural flexibility showcased through fragment-based lead discovery of inhibitors. Nat. Commun. 17, 79 (2026).","DOI":"10.1038\/s41467-025-66768-8"},{"key":"2026093008551220000_2026.09.21.752570v1.16","doi-asserted-by":"crossref","unstructured":"Singh, S. , Clerkin, A. B. , Castellanos, M. A. & Chodera, J. D . More protein-ligand data are needed for AlphaFold-like models to enable drug discovery. Curr. Opin. Struct. Biol. 98, 103257 (2026).","DOI":"10.1016\/j.sbi.2026.103257"},{"key":"2026093008551220000_2026.09.21.752570v1.17","doi-asserted-by":"publisher","DOI":"10.1101\/2025.06.20.660748"},{"key":"2026093008551220000_2026.09.21.752570v1.18","doi-asserted-by":"crossref","unstructured":"Lewis, S. et al. Scalable emulation of protein equilibrium ensembles with generative deep learning. Science (1979). 389, (2025).","DOI":"10.1126\/science.adv9817"},{"key":"2026093008551220000_2026.09.21.752570v1.19","doi-asserted-by":"publisher","DOI":"10.1038\/s41467-024-51507-2"},{"key":"2026093008551220000_2026.09.21.752570v1.20","doi-asserted-by":"crossref","first-page":"735","DOI":"10.1038\/s43588-024-00699-0","article-title":"The future of machine learning for small-molecule drug discovery will be driven by data","volume":"4","year":"2024","journal-title":"Nat. Comput. Sci"}],"container-title":[],"original-title":[],"link":[{"URL":"https:\/\/syndication.highwire.org\/content\/doi\/10.64898\/2026.09.21.752570","content-type":"unspecified","content-version":"vor","intended-application":"similarity-checking"}],"deposited":{"date-parts":[[2026,9,30]],"date-time":"2026-09-30T15:55:25Z","timestamp":1790783725000},"score":1,"resource":{"primary":{"URL":"http:\/\/biorxiv.org\/lookup\/doi\/10.64898\/2026.09.21.752570"}},"subtitle":[],"short-title":[],"issued":{"date-parts":[[2026,9,27]]},"references-count":20,"URL":"https:\/\/doi.org\/10.64898\/2026.09.21.752570","relation":{},"subject":[],"published":{"date-parts":[[2026,9,27]]},"subtype":"preprint"}}