[{"abstract":[{"text":"AlphaFold3 predicts highly accurate protein structures from sequence but tends to collapse to a single dominant conformation, even when the underlying structure is inherently heterogeneous. Moreover, its predictions are oblivious to experimental conditions that can alter local sequence conformation. In this work, we show that AlphaFold3 can be guided to match data obtained by nuclear magnetic resonance (NMR) spectroscopy, X-ray crystallography and cryogenic electron microscopy (cryo-EM) experiments and combinations thereof. Our approach can also incorporate data that explicitly report on dynamics, such as site-resolved order parameters. We demonstrate that this methodology generates compact structural ensembles whose ensemble-averaged observables agree with experiment, with fewer distance restraint violations than traditionally resolved NMR structures and with unmodeled alternate conformations uncovered in electron density. This methodology paves the way for experimentally aware predictive models that generate structural ensembles consistent with the measurements, potentially over multiple modalities, and that can be further refined toward thermodynamically grounded ensembles by incorporating energetics.","lang":"eng"}],"publication_status":"epub_ahead","external_id":{"pmid":["42374114"]},"article_processing_charge":"Yes (via OA deal)","title":"Experiment-guided AlphaFold3 resolves measurement-consistent protein ensembles","OA_place":"publisher","user_id":"ba8df636-2132-11f1-aed0-ed93e2281fdd","author":[{"first_name":"Sai A","last_name":"Maddipatla","id":"e957f5e5-91c9-11f0-a95f-e090f66ecb4d","full_name":"Maddipatla, Sai A"},{"id":"ef280fe0-91c9-11f0-a95f-8dea3f5bc513","full_name":"Sellam, Nadav E","last_name":"Sellam","first_name":"Nadav E"},{"first_name":"Meital I","last_name":"Bojan","id":"11d88cf5-91ca-11f0-a95f-edf9f08f47b7","full_name":"Bojan, Meital I"},{"first_name":"Vova","last_name":"Masalitin","id":"ff7958eb-91c9-11f0-a95f-f3bf65828cf6","full_name":"Masalitin, Vova"},{"full_name":"Vedula, Sanketh","last_name":"Vedula","first_name":"Sanketh"},{"first_name":"Paul","last_name":"Schanda","full_name":"Schanda, Paul","orcid":"0000-0002-9350-7606","id":"7B541462-FAF6-11E9-A490-E8DFE5697425"},{"first_name":"Ailie","last_name":"Marx","full_name":"Marx, Ailie"},{"first_name":"Alexander","last_name":"Bronstein","id":"58f3726e-7cba-11ef-ad8b-e6e8cb3904e6","orcid":"0000-0001-9699-8730","full_name":"Bronstein, Alexander"}],"related_material":{"link":[{"description":"News on ISTA website","relation":"press_release","url":"https://ista.ac.at/en/news/toward-experiment-guided-alphafold/"}]},"pmid":1,"quality_controlled":"1","corr_author":"1","supplementarymaterial":"yes","ddc":["570"],"has_accepted_license":"1","main_file_link":[{"url":"https://doi.org/10.1038/s41587-026-03166-5","open_access":"1"}],"researchdata_availability":"yes","year":"2026","date_created":"2026-07-12T22:02:19Z","publication":"Nature Biotechnology","das_tickbox":"1","publication_identifier":{"issn":["1087-0156"],"eissn":["1546-1696"]},"doi":"10.1038/s41587-026-03166-5","dataavailabilitystatement":"All structures and metrics reported in this paper are openly available on Harvard Dataverse - https://doi.org/10.7910/DVN/PLYUHN. All code is openly available on GitHub (https://github.com/sai-advaith/guided_alphafold); the version used for this paper (version 0.9.1) is permanently archived on Zenodo https://doi.org/10.5281/zenodo.17307005","oa":1,"department":[{"_id":"PaSc"},{"_id":"AlBr"},{"_id":"GradSch"}],"publisher":"Springer Nature","date_updated":"2026-08-04T09:25:18Z","language":[{"iso":"eng"}],"month":"06","date_published":"2026-06-29T00:00:00Z","acknowledgement":"A. Marx acknowledges the financial support of the Helmsley Fellowships Program for Sustainability and Health. A.M.B. and P.S. are supported by the Institute of Science and Technology Austria Internal Project Call grant Generative Protein NMR. S.V. was supported in part by funding from the Eric and Wendy Schmidt Center at the Broad Institute of MIT and Harvard. Open access funding provided by Institute of Science and Technology (IST Austria).","type":"journal_article","article_type":"original","oa_version":"Published Version","_id":"22268","scopus_import":"1","PlanS_conform":"1","citation":{"ama":"Maddipatla SA, Sellam NE, Bojan MI, et al. Experiment-guided AlphaFold3 resolves measurement-consistent protein ensembles. <i>Nature Biotechnology</i>. 2026. doi:<a href=\"https://doi.org/10.1038/s41587-026-03166-5\">10.1038/s41587-026-03166-5</a>","ista":"Maddipatla SA, Sellam NE, Bojan MI, Masalitin V, Vedula S, Schanda P, Marx A, Bronstein AM. 2026. Experiment-guided AlphaFold3 resolves measurement-consistent protein ensembles. Nature Biotechnology.","short":"S.A. Maddipatla, N.E. Sellam, M.I. Bojan, V. Masalitin, S. Vedula, P. Schanda, A. Marx, A.M. Bronstein, Nature Biotechnology (2026).","ieee":"S. A. Maddipatla <i>et al.</i>, “Experiment-guided AlphaFold3 resolves measurement-consistent protein ensembles,” <i>Nature Biotechnology</i>. Springer Nature, 2026.","mla":"Maddipatla, Sai A., et al. “Experiment-Guided AlphaFold3 Resolves Measurement-Consistent Protein Ensembles.” <i>Nature Biotechnology</i>, Springer Nature, 2026, doi:<a href=\"https://doi.org/10.1038/s41587-026-03166-5\">10.1038/s41587-026-03166-5</a>.","apa":"Maddipatla, S. A., Sellam, N. E., Bojan, M. I., Masalitin, V., Vedula, S., Schanda, P., … Bronstein, A. M. (2026). Experiment-guided AlphaFold3 resolves measurement-consistent protein ensembles. <i>Nature Biotechnology</i>. Springer Nature. <a href=\"https://doi.org/10.1038/s41587-026-03166-5\">https://doi.org/10.1038/s41587-026-03166-5</a>","chicago":"Maddipatla, Sai A, Nadav E Sellam, Meital I Bojan, Vova Masalitin, Sanketh Vedula, Paul Schanda, Ailie Marx, and Alex M. Bronstein. “Experiment-Guided AlphaFold3 Resolves Measurement-Consistent Protein Ensembles.” <i>Nature Biotechnology</i>. Springer Nature, 2026. <a href=\"https://doi.org/10.1038/s41587-026-03166-5\">https://doi.org/10.1038/s41587-026-03166-5</a>."},"OA_type":"hybrid","tmp":{"legal_code_url":"https://creativecommons.org/licenses/by/4.0/legalcode","image":"/images/cc_by.png","name":"Creative Commons Attribution 4.0 International Public License (CC-BY 4.0)","short":"CC BY (4.0)"},"status":"public","day":"29"},{"corr_author":"1","supplementarymaterial":"yes","quality_controlled":"1","file":[{"access_level":"open_access","relation":"main_file","checksum":"9f43f5469443388ec55388d95c4cf243","file_size":8534339,"date_updated":"2026-08-18T06:33:14Z","file_id":"22723","success":1,"creator":"dernst","date_created":"2026-08-18T06:33:14Z","file_name":"2026_ICLR_Bojan.pdf","content_type":"application/pdf"}],"has_accepted_license":"1","ddc":["000","570"],"conference":{"start_date":"2026-04-23","name":"ICLR: International Conference on Learning Representations","end_date":"2026-04-27","location":"Rio de Janeiro, Brazil"},"year":"2026","date_created":"2026-08-17T12:03:24Z","researchdata_availability":"yes","dataavailabilitystatement":"The code, trained models, and data-processing scripts are publicly available at https://github.\r\ncom/mb012/MLFF_representation. In addition, complete details of the models and optimization parameters are provided in Appendix G.3. The hardware resources used to produce the\r\nresults are specified in Appendix I.4. The loss functions, evaluation metrics, and details regarding\r\nablation studies are specified in Appendix G. These details ensure that all results reported in the paper\r\ncan be independently verified.","publication":"14th International Conference on Learning Representations","das_tickbox":"1","page":"100760-100799","article_processing_charge":"No","abstract":[{"lang":"eng","text":"The local structure of a protein strongly impacts its function and interactions\r\nwith other molecules. Representing local biomolecular environments remains a\r\nkey challenge while applying machine learning approaches over protein structures. The structural and chemical variability of these environments makes them\r\nchallenging to model, and performing representation learning on these objects\r\nremains largely under-explored. In this work, we propose representations for\r\nlocal protein environments that leverage intermediate features from machine learning force fields (MLFFs). We extensively benchmark state-of-the-art MLFFs,\r\ncomparing their performance across latent spaces and downstream tasks, and\r\nshow that their embeddings capture local structural (e.g., secondary motifs) and\r\nchemical features (e.g., amino acid identity and protonation state), organizing\r\nprotein environments into a structured manifold. We show that these representations enable zero-shot generalization and transfer across diverse downstream\r\ntasks. As a case study, we build a physics-informed, uncertainty-aware chemical shift predictor that achieves state-of-the-art accuracy in biomolecular NMR\r\nspectroscopy. Our results establish MLFFs as general-purpose, reusable representation learners for protein modeling, opening new directions in representation learning for structured physical systems. Code and data are available at\r\nhttps://github.com/mb012/MLFF_representation.\r\n"}],"external_id":{"arxiv":["2505.23354"]},"publication_status":"published","title":"Representing local protein environments with machine learning force fields","author":[{"first_name":"Meital I","last_name":"Bojan","id":"11d88cf5-91ca-11f0-a95f-edf9f08f47b7","full_name":"Bojan, Meital I"},{"first_name":"Sanketh","last_name":"Vedula","full_name":"Vedula, Sanketh"},{"first_name":"Sai A","last_name":"Maddipatla","id":"e957f5e5-91c9-11f0-a95f-e090f66ecb4d","full_name":"Maddipatla, Sai A"},{"last_name":"Sellam","first_name":"Nadav E","id":"ef280fe0-91c9-11f0-a95f-8dea3f5bc513","full_name":"Sellam, Nadav E"},{"first_name":"Anar","last_name":"Rzayev","id":"2cd60677-9acd-11f1-ae1a-a85ae1c4dd35","full_name":"Rzayev, Anar"},{"id":"d42e08e7-f4fc-11eb-af0a-d71e26138f1b","orcid":"0000-0002-9043-136X","full_name":"Napoli, Federico","first_name":"Federico","last_name":"Napoli"},{"last_name":"Schanda","first_name":"Paul","orcid":"0000-0002-9350-7606","id":"7B541462-FAF6-11E9-A490-E8DFE5697425","full_name":"Schanda, Paul"},{"full_name":"Bronstein, Alexander","id":"58f3726e-7cba-11ef-ad8b-e6e8cb3904e6","orcid":"0000-0001-9699-8730","first_name":"Alexander","last_name":"Bronstein"}],"user_id":"2DF688A6-F248-11E8-B48F-1D18A9856A87","volume":2026,"OA_place":"publisher","acknowledged_ssus":[{"_id":"ScienComp"}],"related_material":{"link":[{"url":"https://github.com/mb012/MLFF_representation","relation":"software"}]},"citation":{"chicago":"Bojan, Meital I, Sanketh Vedula, Sai A Maddipatla, Nadav E Sellam, Anar Rzayev, Federico Napoli, Paul Schanda, and Alex M. Bronstein. “Representing Local Protein Environments with Machine Learning Force Fields.” In <i>14th International Conference on Learning Representations</i>, 2026:100760–99, 2026.","apa":"Bojan, M. I., Vedula, S., Maddipatla, S. A., Sellam, N. E., Rzayev, A., Napoli, F., … Bronstein, A. M. (2026). Representing local protein environments with machine learning force fields. In <i>14th International Conference on Learning Representations</i> (Vol. 2026, pp. 100760–100799). Rio de Janeiro, Brazil.","mla":"Bojan, Meital I., et al. “Representing Local Protein Environments with Machine Learning Force Fields.” <i>14th International Conference on Learning Representations</i>, vol. 2026, 2026, pp. 100760–99.","short":"M.I. Bojan, S. Vedula, S.A. Maddipatla, N.E. Sellam, A. Rzayev, F. Napoli, P. Schanda, A.M. Bronstein, in:, 14th International Conference on Learning Representations, 2026, pp. 100760–100799.","ieee":"M. I. Bojan <i>et al.</i>, “Representing local protein environments with machine learning force fields,” in <i>14th International Conference on Learning Representations</i>, Rio de Janeiro, Brazil, 2026, vol. 2026, pp. 100760–100799.","ista":"Bojan MI, Vedula S, Maddipatla SA, Sellam NE, Rzayev A, Napoli F, Schanda P, Bronstein AM. 2026. Representing local protein environments with machine learning force fields. 14th International Conference on Learning Representations. ICLR: International Conference on Learning Representations vol. 2026, 100760–100799.","ama":"Bojan MI, Vedula S, Maddipatla SA, et al. Representing local protein environments with machine learning force fields. In: <i>14th International Conference on Learning Representations</i>. Vol 2026. ; 2026:100760-100799."},"OA_type":"gold","tmp":{"legal_code_url":"https://creativecommons.org/licenses/by/4.0/legalcode","image":"/images/cc_by.png","name":"Creative Commons Attribution 4.0 International Public License (CC-BY 4.0)","short":"CC BY (4.0)"},"day":"01","status":"public","oa":1,"file_date_updated":"2026-08-18T06:33:14Z","month":"05","language":[{"iso":"eng"}],"date_published":"2026-05-01T00:00:00Z","department":[{"_id":"GradSch"},{"_id":"PaSc"},{"_id":"AlBr"}],"arxiv":1,"date_updated":"2026-08-18T06:36:55Z","intvolume":"      2026","oa_version":"Published Version","acknowledgement":"This work was supported by the Institute of Science and Technology Austria (ISTA) through the IPC\r\ngrant “Generative Protein NMR” and by the Israeli Science Foundation (ISF) under grant number\r\n1834/24. This research used resources of the Institute of Science and Technology Austria’s scientific\r\ncomputing cluster. S.V. was supported in part by funding from the Eric and Wendy Schmidt Center at\r\nthe Broad Institute of MIT and Harvard.","type":"conference","_id":"22722"},{"_id":"21327","type":"conference","acknowledgement":"This work was supported by the Israeli Science Foundation (ISF) grant number 1834/24. We acknowledge support from the Austrian Science Fund (FWF, grant numbers I5812-B and I6223) and the financial support of the Helmsley Fellowships Program for Sustainability and Health. This research uses resources of the Institute of Science and Technology Austria’s scientific computing cluster. ","oa_version":"Published Version","intvolume":"       267","department":[{"_id":"PaSc"},{"_id":"AlBr"},{"_id":"GradSch"}],"arxiv":1,"publisher":"ML Research Press","date_updated":"2026-02-19T08:56:43Z","language":[{"iso":"eng"}],"month":"07","date_published":"2025-07-30T00:00:00Z","oa":1,"file_date_updated":"2026-02-19T08:56:10Z","status":"public","day":"30","alternative_title":["PMLR"],"OA_type":"gold","tmp":{"legal_code_url":"https://creativecommons.org/licenses/by/4.0/legalcode","image":"/images/cc_by.png","name":"Creative Commons Attribution 4.0 International Public License (CC-BY 4.0)","short":"CC BY (4.0)"},"project":[{"name":"AlloSpace. The emergence and mechanisms of allostery","grant_number":"I05812","_id":"eb9c82eb-77a9-11ec-83b8-aadd536561cf"},{"name":"Structure and mechanism of the mitochondrial MIM insertase","grant_number":"I06223","_id":"bdb9578d-d553-11ed-ba76-ed5d39fce6f0"}],"citation":{"chicago":"Maddipatla, Sai A, Nadav E Sellam, Meital I Bojan, Sanketh Vedula, Paul Schanda, Ailie Marx, and Alex M. Bronstein. “Inverse Problems with Experiment-Guided AlphaFold.” In <i>Proceedings of the 42nd International Conference on Machine Learning</i>, 267:42366–93. ML Research Press, 2025.","apa":"Maddipatla, S. A., Sellam, N. E., Bojan, M. I., Vedula, S., Schanda, P., Marx, A., &#38; Bronstein, A. M. (2025). Inverse problems with experiment-guided AlphaFold. In <i>Proceedings of the 42nd International Conference on Machine Learning</i> (Vol. 267, pp. 42366–42393). Vancouver, Canada: ML Research Press.","mla":"Maddipatla, Sai A., et al. “Inverse Problems with Experiment-Guided AlphaFold.” <i>Proceedings of the 42nd International Conference on Machine Learning</i>, vol. 267, ML Research Press, 2025, pp. 42366–93.","short":"S.A. Maddipatla, N.E. Sellam, M.I. Bojan, S. Vedula, P. Schanda, A. Marx, A.M. Bronstein, in:, Proceedings of the 42nd International Conference on Machine Learning, ML Research Press, 2025, pp. 42366–42393.","ieee":"S. A. Maddipatla <i>et al.</i>, “Inverse problems with experiment-guided AlphaFold,” in <i>Proceedings of the 42nd International Conference on Machine Learning</i>, Vancouver, Canada, 2025, vol. 267, pp. 42366–42393.","ama":"Maddipatla SA, Sellam NE, Bojan MI, et al. Inverse problems with experiment-guided AlphaFold. In: <i>Proceedings of the 42nd International Conference on Machine Learning</i>. Vol 267. ML Research Press; 2025:42366-42393.","ista":"Maddipatla SA, Sellam NE, Bojan MI, Vedula S, Schanda P, Marx A, Bronstein AM. 2025. Inverse problems with experiment-guided AlphaFold. Proceedings of the 42nd International Conference on Machine Learning. ICML: International Conference on Machine Learning, PMLR, vol. 267, 42366–42393."},"acknowledged_ssus":[{"_id":"ScienComp"}],"volume":267,"OA_place":"publisher","user_id":"2DF688A6-F248-11E8-B48F-1D18A9856A87","author":[{"full_name":"Maddipatla, Sai A","id":"e957f5e5-91c9-11f0-a95f-e090f66ecb4d","last_name":"Maddipatla","first_name":"Sai A"},{"id":"ef280fe0-91c9-11f0-a95f-8dea3f5bc513","full_name":"Sellam, Nadav E","first_name":"Nadav E","last_name":"Sellam"},{"first_name":"Meital I","last_name":"Bojan","full_name":"Bojan, Meital I","id":"11d88cf5-91ca-11f0-a95f-edf9f08f47b7"},{"first_name":"Sanketh","last_name":"Vedula","full_name":"Vedula, Sanketh","id":"94f2fe44-70fa-11f0-b76b-92922c09452b"},{"orcid":"0000-0002-9350-7606","id":"7B541462-FAF6-11E9-A490-E8DFE5697425","full_name":"Schanda, Paul","first_name":"Paul","last_name":"Schanda"},{"last_name":"Marx","first_name":"Ailie","full_name":"Marx, Ailie"},{"first_name":"Alexander","last_name":"Bronstein","id":"58f3726e-7cba-11ef-ad8b-e6e8cb3904e6","orcid":"0000-0001-9699-8730","full_name":"Bronstein, Alexander"}],"title":"Inverse problems with experiment-guided AlphaFold","abstract":[{"text":"Proteins exist as a dynamic ensemble of multiple conformations, and these motions are often crucial for their functions. However, current structure prediction methods predominantly yield a single conformation, overlooking the conformational heterogeneity revealed by diverse experimental modalities. Here, we present a framework for building experiment-grounded protein structure generative models that infer conformational ensembles consistent with measured experimental data. The key idea is to treat stateof-the-art protein structure predictors (e.g., AlphaFold3) as sequence-conditioned structural priors, and cast ensemble modeling as posterior inference of protein structures given experimental measurements. Through extensive real-data experiments, we demonstrate the generality of our method to incorporate a variety of experimental measurements. In particular, our framework uncovers previously unmodeled conformational heterogeneity from crystallographic densities, and generates high-accuracy NMR ensembles orders of magnitude faster than the status quo. Notably, we demonstrate that our ensembles outperform AlphaFold3 (Abramson et al., 2024) and sometimes better fit experimental data than publicly deposited structures to the Protein Data Bank (PDB, Burley et al. (2017)). We believe that this approach will unlock building predictive models that fully embrace experimentally observed conformational diversity.","lang":"eng"}],"external_id":{"arxiv":["2502.09372"]},"publication_status":"published","page":"42366 - 42393","article_processing_charge":"No","publication":"Proceedings of the 42nd International Conference on Machine Learning","publication_identifier":{"eissn":["2640-3498"]},"year":"2025","date_created":"2026-02-18T12:11:17Z","ddc":["000","540"],"conference":{"start_date":"2025-07-13","name":"ICML: International Conference on Machine Learning","end_date":"2025-07-19","location":"Vancouver, Canada"},"has_accepted_license":"1","file":[{"creator":"dernst","success":1,"content_type":"application/pdf","file_name":"2025_ICML_Maddipatla.pdf","date_created":"2026-02-19T08:56:10Z","date_updated":"2026-02-19T08:56:10Z","file_size":1924177,"checksum":"f33230a6d59b7978d4cd72795e4e9059","relation":"main_file","access_level":"open_access","file_id":"21338"}],"quality_controlled":"1","corr_author":"1"}]