{"status":"ok","message-type":"work","message-version":"1.0.0","message":{"institution":[{"name":"bioRxiv"}],"indexed":{"date-parts":[[2026,9,23]],"date-time":"2026-09-23T22:19:41Z","timestamp":1790201981518,"version":"4.1.0"},"posted":{"date-parts":[[2024,8,26]]},"group-title":"Biophysics","reference-count":54,"publisher":"openRxiv","license":[{"start":{"date-parts":[[2024,8,26]],"date-time":"2024-08-26T00:00:00Z","timestamp":1724630400000},"content-version":"vor","delay-in-days":0,"URL":"http:\/\/creativecommons.org\/licenses\/by-nc-nd\/4.0\/"}],"content-domain":{"domain":[],"crossmark-restriction":false},"short-container-title":[],"accepted":{"date-parts":[[2024,8,26]]},"abstract":"<jats:p>\n                  AlphaFold2 (AF2) has revolutionized protein structure prediction. However, a common confusion lies in equating the\n                  <jats:italic>protein structure prediction<\/jats:italic>\n                  problem with the\n                  <jats:italic>protein folding problem<\/jats:italic>\n                  . The former provides a static structure, while the latter explains the dynamic folding pathway to that structure. We challenge the current\n                  <jats:italic>status quo<\/jats:italic>\n                  and advocate that AF2 has indeed learned some protein folding prin- ciples, despite being designed for structure prediction. AF2\u2019s high-dimensional parameters encode an imperfect biophysical scoring function. Typically, AF2 uses multiple sequence alignments (MSAs) to guide the search within a narrow re- gion of its learned surface. In our study, we operate AF2 without MSAs or initial templates, forcing it to sample its entire energy landscape \u2014 more akin to an\n                  <jats:italic>ab initio<\/jats:italic>\n                  approach. Among over 7,000 proteins, a fraction fold using sequence alone, highlighting the smoothness of AF2\u2019s learned energy surface. Additionally, by combining recycling and iterative predictions, we discover multiple AF2 interme- diate structures in good agreement with known experimental data. AF2 appears to follow a \u201clocal first, global later\u201d folding mechanism. For designed proteins with more optimized local interactions, AF2\u2019s energy landscape is too smooth to detect intermediates even when it should. Our current work sheds new light on what AF2 has learned and opens exciting possibilities to advance our understanding of protein folding and for experimental discovery of folding intermediates.\n                <\/jats:p>","DOI":"10.1101\/2024.08.25.609581","type":"posted-content","created":{"date-parts":[[2024,8,29]],"date-time":"2024-08-29T05:29:25Z","timestamp":1724909365000},"source":"Crossref","is-referenced-by-count":6,"title":["AlphaFold2\n                  <i>knows<\/i>\n                  some protein folding principles"],"prefix":"10.64898","author":[{"ORCID":"https:\/\/orcid.org\/0000-0001-5847-0820","authenticated-orcid":false,"given":"Liwei","family":"Chang","sequence":"first","affiliation":[],"role":[{"vocabulary":"crossref","role":"author"}]},{"ORCID":"https:\/\/orcid.org\/0000-0002-5054-5338","authenticated-orcid":false,"given":"Alberto","family":"Perez","sequence":"additional","affiliation":[],"role":[{"vocabulary":"crossref","role":"author"}]}],"member":"54368","reference":[{"key":"2024082902101020000_2024.08.25.609581v1.1","doi-asserted-by":"crossref","unstructured":"J. Jumper , et al., Highly accurate protein structure prediction with AlphaFold. Nature pp. 1\u201311 (2021).","DOI":"10.1038\/s41592-021-01362-6"},{"key":"2024082902101020000_2024.08.25.609581v1.2","unstructured":"R. Evans , et al., Protein complex prediction with AlphaFold-Multimer. bioRxiv p. 2021.10.04.463034 (2021)."},{"key":"2024082902101020000_2024.08.25.609581v1.3","doi-asserted-by":"crossref","unstructured":"T. Tsaban , et al., Harnessing protein folding neural networks for peptide\u2013protein docking. Nature Communications 13 (1), 176 (2022).","DOI":"10.1038\/s41467-021-27838-9"},{"key":"2024082902101020000_2024.08.25.609581v1.4","doi-asserted-by":"crossref","unstructured":"L. Chang , A. Perez , Ranking Peptide Binders by Affinity with AlphaFold. Angewandte Chemie International Edition (2022).","DOI":"10.1002\/anie.202213362"},{"key":"2024082902101020000_2024.08.25.609581v1.5","doi-asserted-by":"crossref","unstructured":"D. d. Alamo , D. Sala , H. S. Mchaourab , J. Meiler , Sampling alternative conformational states of transporters and receptors with AlphaFold2. eLife 11 (2022).","DOI":"10.7554\/eLife.75751"},{"key":"2024082902101020000_2024.08.25.609581v1.6","doi-asserted-by":"crossref","unstructured":"H. K. Wayment-Steele , et al., Predicting multiple conformations via sequence clustering and AlphaFold2. Nature pp. 1\u20133 (2023).","DOI":"10.1101\/2022.10.17.512570"},{"issue":"1","key":"2024082902101020000_2024.08.25.609581v1.7","doi-asserted-by":"crossref","first-page":"5478","DOI":"10.1038\/s41467-023-41237-2","article-title":"Evolutionary selection of proteins with two folds","volume":"14","year":"2023","journal-title":"Nature Commu- nications"},{"key":"2024082902101020000_2024.08.25.609581v1.8","doi-asserted-by":"crossref","unstructured":"A. Elofsson , Progress at protein structure prediction, as seen in CASP15. Current Opinion in Structural Biology 80, 102594 (2023).","DOI":"10.1016\/j.sbi.2023.102594"},{"issue":"1","key":"2024082902101020000_2024.08.25.609581v1.9","doi-asserted-by":"crossref","first-page":"e2214423119","DOI":"10.1073\/pnas.2214423119","article-title":"Protein folds vs. protein folding: Differing questions, different challenges","volume":"120","year":"2023","journal-title":"Proceedings of the National Academy of Sciences"},{"issue":"6574","key":"2024082902101020000_2024.08.25.609581v1.10","doi-asserted-by":"crossref","first-page":"1415","DOI":"10.1126\/science.abn5795","article-title":"Proteins, proteins everywhere","volume":"374","year":"2021","journal-title":"Science"},{"key":"2024082902101020000_2024.08.25.609581v1.11","doi-asserted-by":"publisher","DOI":"10.1126\/science.abn9422"},{"key":"2024082902101020000_2024.08.25.609581v1.12","doi-asserted-by":"crossref","unstructured":"J. P. Roney , S. Ovchinnikov , State-of-the-Art Estimation of Protein Model Accuracy Using AlphaFold. Physical Review Letters 129 (23), 238101 (2022).","DOI":"10.1103\/PhysRevLett.129.238101"},{"key":"2024082902101020000_2024.08.25.609581v1.13","doi-asserted-by":"crossref","unstructured":"L. Chang , A. Perez , Deciphering the Folding Mechanism of Proteins G and L and Their Mutants. Journal of the American Chemical Society (2022).","DOI":"10.1021\/jacs.2c04488"},{"key":"2024082902101020000_2024.08.25.609581v1.14","doi-asserted-by":"publisher","DOI":"10.1093\/nar\/28.1.235"},{"key":"2024082902101020000_2024.08.25.609581v1.15","doi-asserted-by":"publisher","DOI":"10.1038\/89638"},{"key":"2024082902101020000_2024.08.25.609581v1.16","doi-asserted-by":"publisher","DOI":"10.1110\/ps.0216902"},{"key":"2024082902101020000_2024.08.25.609581v1.17","doi-asserted-by":"publisher","DOI":"10.1073\/pnas.1503613112"},{"key":"2024082902101020000_2024.08.25.609581v1.18","doi-asserted-by":"publisher","DOI":"10.1006\/jmbi.2000.3701"},{"key":"2024082902101020000_2024.08.25.609581v1.19","doi-asserted-by":"publisher","DOI":"10.1016\/j.jmb.2012.04.013"},{"key":"2024082902101020000_2024.08.25.609581v1.20","doi-asserted-by":"publisher","DOI":"10.1006\/jmbi.2001.5229"},{"key":"2024082902101020000_2024.08.25.609581v1.21","doi-asserted-by":"publisher","DOI":"10.1126\/science.1208351"},{"key":"2024082902101020000_2024.08.25.609581v1.22","doi-asserted-by":"publisher","DOI":"10.1016\/j.bpj.2014.06.037"},{"key":"2024082902101020000_2024.08.25.609581v1.23","doi-asserted-by":"publisher","DOI":"10.1021\/jacs.5b11300"},{"key":"2024082902101020000_2024.08.25.609581v1.24","doi-asserted-by":"crossref","unstructured":"H. Maity , G. Reddy , Transient intermediates are populated in the folding pathways of single- domain two-state folding protein L. Journal of Chemical Physics 148 (16), 165101 (2018).","DOI":"10.1063\/1.5022632"},{"key":"2024082902101020000_2024.08.25.609581v1.25","doi-asserted-by":"crossref","unstructured":"A. Mitsutake , H. Takano , Folding pathways of NuG2\u2014a designed mutant of protein G\u2014using relaxation mode analysis. Journal of Chemical Physics 151 (4), 044117 (2019).","DOI":"10.1063\/1.5097708"},{"issue":"11","key":"2024082902101020000_2024.08.25.609581v1.26","doi-asserted-by":"crossref","first-page":"e1008323","DOI":"10.1371\/journal.pcbi.1008323","article-title":"Validation of DBFOLD: An efficient algorithm for computing folding pathways of complex proteins","volume":"16","year":"2020","journal-title":"PLoS Computational Biology"},{"key":"2024082902101020000_2024.08.25.609581v1.27","doi-asserted-by":"publisher","DOI":"10.1073\/pnas.0407683101"},{"key":"2024082902101020000_2024.08.25.609581v1.28","doi-asserted-by":"publisher","DOI":"10.1093\/protein\/gzi025"},{"key":"2024082902101020000_2024.08.25.609581v1.29","doi-asserted-by":"publisher","DOI":"10.1016\/j.jmb.2004.01.018"},{"key":"2024082902101020000_2024.08.25.609581v1.30","doi-asserted-by":"publisher","DOI":"10.1016\/j.jmb.2008.01.012"},{"issue":"15","key":"2024082902101020000_2024.08.25.609581v1.31","doi-asserted-by":"crossref","first-page":"5915","DOI":"10.1073\/pnas.1218321110","article-title":"Atomic-level description of ubiquitin folding","volume":"110","year":"2013","journal-title":"Pro- ceedings of the National Academy of Sciences"},{"key":"2024082902101020000_2024.08.25.609581v1.32","doi-asserted-by":"publisher","DOI":"10.1073\/pnas.97.13.7084"},{"key":"2024082902101020000_2024.08.25.609581v1.33","doi-asserted-by":"publisher","DOI":"10.1038\/14901"},{"key":"2024082902101020000_2024.08.25.609581v1.34","doi-asserted-by":"publisher","DOI":"10.1126\/science.1214203"},{"key":"2024082902101020000_2024.08.25.609581v1.35","doi-asserted-by":"crossref","first-page":"44","DOI":"10.1051\/jcp\/1968650044","article-title":"Are there pathways for protein folding?","volume":"65","year":"1968","journal-title":"Journal de chimie physique"},{"key":"2024082902101020000_2024.08.25.609581v1.36","doi-asserted-by":"publisher","DOI":"10.1021\/bi00327a032"},{"key":"2024082902101020000_2024.08.25.609581v1.37","doi-asserted-by":"publisher","DOI":"10.1073\/pnas.89.18.8721"},{"key":"2024082902101020000_2024.08.25.609581v1.38","doi-asserted-by":"crossref","unstructured":"J. Dauparas , et al., Robust deep learning\u2013based protein sequence design using ProteinMPNN. Science (2022).","DOI":"10.1101\/2022.06.03.494563"},{"issue":"4","key":"2024082902101020000_2024.08.25.609581v1.39","doi-asserted-by":"crossref","first-page":"985","DOI":"10.1006\/jmbi.1998.1645","article-title":"Contact order, transition state placement and the refolding rates of single domain proteins11Edited by P","volume":"277","year":"1998","journal-title":"E. Wright. Journal of Molecular Biology"},{"key":"2024082902101020000_2024.08.25.609581v1.40","doi-asserted-by":"publisher","DOI":"10.1110\/ps.03403604"},{"key":"2024082902101020000_2024.08.25.609581v1.41","doi-asserted-by":"publisher","DOI":"10.1006\/jmbi.2001.4775"},{"key":"2024082902101020000_2024.08.25.609581v1.42","doi-asserted-by":"publisher","DOI":"10.1110\/ps.034660.108"},{"key":"2024082902101020000_2024.08.25.609581v1.43","doi-asserted-by":"publisher","DOI":"10.1073\/pnas.2636460100"},{"key":"2024082902101020000_2024.08.25.609581v1.44","doi-asserted-by":"publisher","DOI":"10.1016\/S0092-8674(00)81002-8"},{"key":"2024082902101020000_2024.08.25.609581v1.45","doi-asserted-by":"publisher","DOI":"10.1006\/jmbi.1998.1863"},{"issue":"5","key":"2024082902101020000_2024.08.25.609581v1.46","doi-asserted-by":"crossref","first-page":"763","DOI":"10.1006\/jmbi.1997.1148","article-title":"A comparison of the folding kinetics and thermodynamics of two homologous fibronectin type III modules11Edited by P","volume":"270","year":"1997","journal-title":"E. Wright. Journal of Molecular Biology"},{"key":"2024082902101020000_2024.08.25.609581v1.47","doi-asserted-by":"publisher","DOI":"10.1016\/S0022-2836(02)00184-5"},{"key":"2024082902101020000_2024.08.25.609581v1.48","doi-asserted-by":"crossref","unstructured":"Y. Su , R. E. Iacob , J. Li , J. R. Engen , T. A. Springer , Dynamics of integrin \u03b15\u03b21, fibronectin, and their complex reveal sites of interaction and conformational change. Journal of Biological Chemistry 298 (9), 102323 (2022).","DOI":"10.1016\/j.jbc.2022.102323"},{"key":"2024082902101020000_2024.08.25.609581v1.49","doi-asserted-by":"publisher","DOI":"10.1073\/pnas.1506788112"},{"issue":"7","key":"2024082902101020000_2024.08.25.609581v1.50","doi-asserted-by":"crossref","first-page":"665","DOI":"10.1038\/s41592-020-0848-2","article-title":"Macromolecular modeling and design in Rosetta: recent methods and frameworks","volume":"17","year":"2020","journal-title":"Nature Methods"},{"key":"2024082902101020000_2024.08.25.609581v1.51","doi-asserted-by":"publisher","DOI":"10.1021\/jacs.8b10735"},{"key":"2024082902101020000_2024.08.25.609581v1.52","doi-asserted-by":"publisher","DOI":"10.1021\/acs.jcim.3c01318"},{"issue":"W1","key":"2024082902101020000_2024.08.25.609581v1.53","doi-asserted-by":"crossref","first-page":"W304","DOI":"10.1093\/nar\/gky328","article-title":"UNRES server for physics-based coarse-grained simulations and prediction of protein structure, dynamics and thermodynamics","volume":"46","year":"2018","journal-title":"Nucleic Acids Research"},{"key":"2024082902101020000_2024.08.25.609581v1.54","doi-asserted-by":"publisher","DOI":"10.1063\/1.464068"}],"container-title":[],"original-title":[],"link":[{"URL":"https:\/\/syndication.highwire.org\/content\/doi\/10.1101\/2024.08.25.609581","content-type":"unspecified","content-version":"vor","intended-application":"similarity-checking"}],"deposited":{"date-parts":[[2026,1,15]],"date-time":"2026-01-15T06:56:24Z","timestamp":1768460184000},"score":1,"resource":{"primary":{"URL":"http:\/\/biorxiv.org\/lookup\/doi\/10.1101\/2024.08.25.609581"}},"subtitle":[],"short-title":[],"issued":{"date-parts":[[2024,8,26]]},"references-count":54,"URL":"https:\/\/doi.org\/10.1101\/2024.08.25.609581","relation":{},"subject":[],"published":{"date-parts":[[2024,8,26]]},"subtype":"preprint"}}