Abstract: Proteins are the essential parts of living organisms and they participate in virtually every process within cells. An in-depth understanding of the spatiotemporal multi-scale structure and the dominating mechanisms of protein structures would be the basis for scientific exploration of the origin of life, the mechanisms of diseases and the development of new drugs. Due to the limitations of the spatial and temporal resolutions of current experimental methods, computer simulations, especially molecular dynamics simulations, have become one of the most important methods to study the structure and function of protein systems. This article reviewed the progress of molecular simulations and their application in the research of protein systems during the past half century, especially for molecular dynamics simulations and enhanced sampling methods. The time and space involved in protein simulations covers a wide range of scales, which makes it a great challenge to simulate the spatial-temporal multi-scale structures quickly and accurately, or to investigate the physiological process and the underlying dominating mechanisms. Therefore, this article summarized the recent development of the theoretical models and computing algorithms, and their applications in the investigations of the molecular mechanisms of the native structures and structural changes of the structured proteins, the dynamic structure ensemble of intrinsic disordered proteins and the coupled folding and binding with target protein or other biological molecules, protein complex such as molecular chaperonin, virus particle, etc. Furthermore, the evolution of the popular softwares for molecular dynamics simulations driven by the rapid development of high-performance super computers, and their acceleration of the spatial-temporal scales in molecular dynamics simulations of protein systems, were further discussed. At the end of the article, based on the rapid development of mesoscience theory and its successful applications in a variety of complex systems, the future simulation methods and theoretical research of protein systems were prospected.

Key words: Protein, molecular dynamics simulations, spatial-temporal multi-scale structures, mesoscience

摘要： 蛋白质是生命的物质基础，是生命活动的主要承担者，对蛋白质时空多尺度结构及其控制机制的深入理解是探索生命起源、病理认知及新药开发的基础. 受实验表征手段及时空分辨率的限制，计算机模拟已成为研究蛋白质体系结构及功能的重要手段之一. 由于蛋白质体系模拟所涉及的时间和空间跨度均相当大，因此，准确且快速地描述其时空多尺度结构，从而分析体系的控制机制及相关生理过程，成为分子动力学模拟面临的巨大挑战. 本工作对近半个世纪以来的分子模拟方法，特别是分子动力学方法和相关的增强采样技术在蛋白质体系研究中的应用进行了总结，综述了近年来分子动力学的理论模型和算法的发展，并介绍了这些方法在结构化蛋白质的天然结构与构象变化、固有无序蛋白质的动态结构及其结合底物的动力学过程及分子机理、分子伴侣及病毒等蛋白质复合物体系中的研究成果；汇总了高性能计算的飞速发展所带动的分子动力学模拟软件的变革，拓展了蛋白质模拟的时空尺度，重点阐述了大规模高性能分子动力学模拟在蛋白质研究中的应用；最后，基于介科学理论的飞速发展及其在多种复杂体系的成功运用，对未来蛋白质体系的模拟方法和理论研究的趋势进行了思考和展望.

关键词: 蛋白质, 分子动力学模拟, 时空多尺度结构, 介科学