{"title":"The power and pitfalls of AlphaFold2 for structure prediction beyond rigid globular proteins","authors":"Vinayak Agarwal, Andrew C. McShan","doi":"10.1038/s41589-024-01638-w","DOIUrl":null,"url":null,"abstract":"Artificial intelligence-driven advances in protein structure prediction in recent years have raised the question: has the protein structure-prediction problem been solved? Here, with a focus on nonglobular proteins, we highlight the many strengths and potential weaknesses of DeepMind’s AlphaFold2 in the context of its biological and therapeutic applications. We summarize the subtleties associated with evaluation of AlphaFold2 model quality and reliability using the predicted local distance difference test (pLDDT) and predicted aligned error (PAE) values. We highlight various classes of proteins that AlphaFold2 can be applied to and the caveats involved. Concrete examples of how AlphaFold2 models can be integrated with experimental data in the form of small-angle X-ray scattering (SAXS), solution NMR, cryo-electron microscopy (cryo-EM) and X-ray diffraction are discussed. Finally, we highlight the need to move beyond structure prediction of rigid, static structural snapshots toward conformational ensembles and alternate biologically relevant states. The overarching theme is that careful consideration is due when using AlphaFold2-generated models to generate testable hypotheses and structural models, rather than treating predicted models as de facto ground truth structures. This Perspective proposes practical guidance to the application of AlphaFold2 for structure prediction of different classes of proteins including rigid globular proteins, intrinsically disordered proteins and alternative conformational states. The use of evaluation metrics to predict reliability of the resulting models and their integration with experimental data are also discussed.","PeriodicalId":18832,"journal":{"name":"Nature chemical biology","volume":null,"pages":null},"PeriodicalIF":12.9000,"publicationDate":"2024-06-21","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":"0","resultStr":null,"platform":"Semanticscholar","paperid":null,"PeriodicalName":"Nature chemical biology","FirstCategoryId":"99","ListUrlMain":"https://www.nature.com/articles/s41589-024-01638-w","RegionNum":1,"RegionCategory":"生物学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":null,"EPubDate":"","PubModel":"","JCR":"Q1","JCRName":"BIOCHEMISTRY & MOLECULAR BIOLOGY","Score":null,"Total":0}
引用次数: 0
Abstract
Artificial intelligence-driven advances in protein structure prediction in recent years have raised the question: has the protein structure-prediction problem been solved? Here, with a focus on nonglobular proteins, we highlight the many strengths and potential weaknesses of DeepMind’s AlphaFold2 in the context of its biological and therapeutic applications. We summarize the subtleties associated with evaluation of AlphaFold2 model quality and reliability using the predicted local distance difference test (pLDDT) and predicted aligned error (PAE) values. We highlight various classes of proteins that AlphaFold2 can be applied to and the caveats involved. Concrete examples of how AlphaFold2 models can be integrated with experimental data in the form of small-angle X-ray scattering (SAXS), solution NMR, cryo-electron microscopy (cryo-EM) and X-ray diffraction are discussed. Finally, we highlight the need to move beyond structure prediction of rigid, static structural snapshots toward conformational ensembles and alternate biologically relevant states. The overarching theme is that careful consideration is due when using AlphaFold2-generated models to generate testable hypotheses and structural models, rather than treating predicted models as de facto ground truth structures. This Perspective proposes practical guidance to the application of AlphaFold2 for structure prediction of different classes of proteins including rigid globular proteins, intrinsically disordered proteins and alternative conformational states. The use of evaluation metrics to predict reliability of the resulting models and their integration with experimental data are also discussed.
期刊介绍:
Nature Chemical Biology stands as an esteemed international monthly journal, offering a prominent platform for the chemical biology community to showcase top-tier original research and commentary. Operating at the crossroads of chemistry, biology, and related disciplines, chemical biology utilizes scientific ideas and approaches to comprehend and manipulate biological systems with molecular precision.
The journal embraces contributions from the growing community of chemical biologists, encompassing insights from chemists applying principles and tools to biological inquiries and biologists striving to comprehend and control molecular-level biological processes. We prioritize studies unveiling significant conceptual or practical advancements in areas where chemistry and biology intersect, emphasizing basic research, especially those reporting novel chemical or biological tools and offering profound molecular-level insights into underlying biological mechanisms.
Nature Chemical Biology also welcomes manuscripts describing applied molecular studies at the chemistry-biology interface due to the broad utility of chemical biology approaches in manipulating or engineering biological systems. Irrespective of scientific focus, we actively seek submissions that creatively blend chemistry and biology, particularly those providing substantial conceptual or methodological breakthroughs with the potential to open innovative research avenues. The journal maintains a robust and impartial review process, emphasizing thorough chemical and biological characterization.