AlphaFold 2：为何它能奏效，及其对理解蛋白质序列、结构和功能关系的启示。
AlphaFold 2: Why It Works and Its Implications for Understanding the Relationships of Protein Sequence, Structure, and Function.
机构信息
Center for the Study of Systems Biology, School of Biological Sciences, Georgia Institute of Technology, Atlanta, Georgia 30332, United States.
Twilight Design, 4 Adams Road, Kendall Park, New Jersey 08824, United States.
出版信息
J Chem Inf Model. 2021 Oct 25;61(10):4827-4831. doi: 10.1021/acs.jcim.1c01114. Epub 2021 Sep 29.
Abstract
AlphaFold 2 (AF2) was the star of CASP14, the last biannual structure prediction experiment. Using novel deep learning, AF2 predicted the structures of many difficult protein targets at or near experimental resolution. Here, we present our perspective of why AF2 works and show that it is a very sophisticated fold recognition algorithm that exploits the completeness of the library of single domain PDB structures. It has also learned local side chain packing rearrangements that enable it to refine proteins to high resolution. The benefits and limitations of its ability to predict the structures of many more proteins at or close to atomic detail are discussed.
摘要
AlphaFold 2（AF2）是 CASP14 的明星，这是最后一次两年一次的结构预测实验。使用新的深度学习，AF2 预测了许多具有实验分辨率或接近实验分辨率的困难蛋白靶标的结构。在这里，我们提出了为什么 AF2 起作用的观点，并表明它是一种非常复杂的折叠识别算法，利用了单域 PDB 结构库的完整性。它还学习了局部侧链包装的重新排列，使其能够将蛋白质精修到高分辨率。讨论了它能够预测更多具有原子细节的蛋白质结构的能力的优势和局限性。
相似文献
1
AlphaFold 2: Why It Works and Its Implications for Understanding the Relationships of Protein Sequence, Structure, and Function.AlphaFold 2：为何它能奏效，及其对理解蛋白质序列、结构和功能关系的启示。
J Chem Inf Model. 2021 Oct 25;61(10):4827-4831. doi: 10.1021/acs.jcim.1c01114. Epub 2021 Sep 29.
2
Applying and improving AlphaFold at CASP14.应用和改进 AlphaFold 参加 CASP14。
Proteins. 2021 Dec;89(12):1711-1721. doi: 10.1002/prot.26257.
3
Using AlphaFold to predict the impact of single mutations on protein stability and function.利用 AlphaFold 预测单突变对蛋白质稳定性和功能的影响。
PLoS One. 2023 Mar 16;18(3):e0282689. doi: 10.1371/journal.pone.0282689. eCollection 2023.
4
AlphaFold and Implications for Intrinsically Disordered Proteins.AlphaFold 及其对无序蛋白质的影响。
J Mol Biol. 2021 Oct 1;433(20):167208. doi: 10.1016/j.jmb.2021.167208. Epub 2021 Aug 18.
5
De novo protein design by inversion of the AlphaFold structure prediction network.通过反转 AlphaFold 结构预测网络进行从头设计蛋白质。
Protein Sci. 2023 Jun;32(6):e4653. doi: 10.1002/pro.4653.
6
The AlphaFold Database of Protein Structures: A Biologist's Guide.蛋白质结构的AlphaFold数据库：生物学家指南
J Mol Biol. 2022 Jan 30;434(2):167336. doi: 10.1016/j.jmb.2021.167336. Epub 2021 Oct 29.
7
Understanding the general packing rearrangements required for successful template based modeling of protein structure from a CASP experiment.理解基于模板的蛋白质结构建模从 CASP 实验中成功所需的一般包装重排。
Comput Biol Chem. 2013 Feb;42:40-8. doi: 10.1016/j.compbiolchem.2012.10.008. Epub 2012 Nov 23.
8
Collective Variable for Metadynamics Derived From AlphaFold Output.源自AlphaFold输出的元动力学集体变量。
Front Mol Biosci. 2022 Jun 13;9:878133. doi: 10.3389/fmolb.2022.878133. eCollection 2022.
9
Prediction of protein local structures and folding fragments based on building-block library.基于构建模块库的蛋白质局部结构和折叠片段预测
Proteins. 2008 Jul;72(1):353-66. doi: 10.1002/prot.21931.
10
AlphaFold predictions of fold-switched conformations are driven by structure memorization.AlphaFold 对构象转换构象的预测是由结构记忆驱动的。
Nat Commun. 2024 Aug 24;15(1):7296. doi: 10.1038/s41467-024-51801-z.
引用本文的文献
1
Multiple myeloma associated long non-coding RNA PLUM confers chemoresistance by enhancing PRC2 mediated UPR pathway activation.多发性骨髓瘤相关长链非编码RNA PLUM通过增强PRC2介导的未折叠蛋白反应（UPR）途径激活赋予化疗耐药性。
Nat Commun. 2025 Sep 1;16(1):8155. doi: 10.1038/s41467-025-63256-x.
2
AI-driven epitope prediction: a system review, comparative analysis, and practical guide for vaccine development.人工智能驱动的表位预测：疫苗开发的系统综述、比较分析及实用指南
NPJ Vaccines. 2025 Aug 30;10(1):207. doi: 10.1038/s41541-025-01258-y.
3
High-resolution Cryo-EM Analysis of the Therapeutic Pseudomonas Phage Pa223.治疗性假单胞菌噬菌体Pa223的高分辨率冷冻电镜分析
J Mol Biol. 2025 Aug 12;437(21):169386. doi: 10.1016/j.jmb.2025.169386.
4
AF3Complex yields improved structural predictions of protein complexes.AF3复合物能提高蛋白质复合物的结构预测能力。
Bioinformatics. 2025 Aug 2;41(8). doi: 10.1093/bioinformatics/btaf432.
5
Structural insights into Beclin 1 interactions with it's regulators for autophagy modulation.对Beclin 1与其自噬调节因子相互作用的结构见解。
Comput Struct Biotechnol J. 2025 Jul 7;27:3005-3035. doi: 10.1016/j.csbj.2025.06.044. eCollection 2025.
6
Neuropsychopharmacology in the era of artificial intelligence and biomolecule prediction software.人工智能与生物分子预测软件时代的神经精神药理学
NPP Digit Psychiatry Neurosci. 2025;3(1):16. doi: 10.1038/s44277-025-00038-9. Epub 2025 Jun 30.
7
Regulation and induction of fungal secondary metabolites: a comprehensive review.真菌次生代谢产物的调控与诱导：综述
Arch Microbiol. 2025 Jul 1;207(8):189. doi: 10.1007/s00203-025-04386-0.
8
Toward streamline variant classification: discrepancies in variant nomenclature and syntax for ClinVar pathogenic variants across annotation tools.迈向简化变异分类：跨注释工具的ClinVar致病变异的变异命名法和语法差异
Hum Genomics. 2025 Jun 21;19(1):70. doi: 10.1186/s40246-025-00778-x.
9
Multimeric protein interaction and complex prediction: Structure, dynamics and function.多聚体蛋白质相互作用与复合物预测：结构、动力学与功能
Comput Struct Biotechnol J. 2025 May 16;27:1975-1997. doi: 10.1016/j.csbj.2025.05.009. eCollection 2025.
10
Cutting-edge deep-learning based tools for metagenomic research.用于宏基因组学研究的前沿深度学习工具。
Natl Sci Rev. 2025 Feb 19;12(6):nwaf056. doi: 10.1093/nsr/nwaf056. eCollection 2025 Jun.
本文引用的文献
1
Improved protein structure prediction by deep learning irrespective of co-evolution information.通过深度学习改进蛋白质结构预测，与共进化信息无关。
Nat Mach Intell. 2021 Jul;3:601-609. doi: 10.1038/s42256-021-00348-5. Epub 2021 May 20.
2
Protein tertiary structure prediction and refinement using deep learning and Rosetta in CASP14.使用深度学习和 Rosetta 在 CASP14 中进行蛋白质三级结构预测和精修。
Proteins. 2021 Dec;89(12):1722-1733. doi: 10.1002/prot.26194. Epub 2021 Aug 17.
3
A General Framework to Learn Tertiary Structure for Protein Sequence Characterization.用于蛋白质序列特征描述的三级结构学习通用框架。
Front Bioinform. 2021 May;1. doi: 10.3389/fbinf.2021.689960. Epub 2021 May 21.
4
Highly accurate protein structure prediction for the human proteome.高精准度的人类蛋白质组蛋白结构预测。
Nature. 2021 Aug;596(7873):590-596. doi: 10.1038/s41586-021-03828-1. Epub 2021 Jul 22.
5
Improving protein tertiary structure prediction by deep learning and distance prediction in CASP14.通过深度学习和距离预测改进 CASP14 中的蛋白质三级结构预测。
Proteins. 2022 Jan;90(1):58-72. doi: 10.1002/prot.26186. Epub 2021 Jul 27.
6
Accurate prediction of protein structures and interactions using a three-track neural network.使用三轨神经网络准确预测蛋白质结构和相互作用。
Science. 2021 Aug 20;373(6557):871-876. doi: 10.1126/science.abj8754. Epub 2021 Jul 15.
7
Highly accurate protein structure prediction with AlphaFold.利用 AlphaFold 进行高精度蛋白质结构预测。
Nature. 2021 Aug;596(7873):583-589. doi: 10.1038/s41586-021-03819-2. Epub 2021 Jul 15.
8
Binding mechanism of inhibitors to p38α MAP kinase deciphered by using multiple replica Gaussian accelerated molecular dynamics and calculations of binding free energies.通过使用多重复制高斯加速分子动力学和结合自由能计算，揭示了抑制剂与 p38α MAP 激酶的结合机制。
Comput Biol Med. 2021 Jul;134:104485. doi: 10.1016/j.compbiomed.2021.104485. Epub 2021 May 11.
9
FRAGSITE: A Fragment-Based Approach for Virtual Ligand Screening.FRAGSITE：基于片段的虚拟配体筛选方法。
J Chem Inf Model. 2021 Apr 26;61(4):2074-2089. doi: 10.1021/acs.jcim.0c01160. Epub 2021 Mar 16.
10
Combination of deep neural network with attention mechanism enhances the explainability of protein contact prediction.深度神经网络与注意力机制的结合增强了蛋白质接触预测的可解释性。
Proteins. 2021 Jun;89(6):697-707. doi: 10.1002/prot.26052. Epub 2021 Feb 16.
Zotero 插件