Free Energy Sampling of Long-Timescale Biomolecular Dynamics

项目概要 蛋白质在其周围无目的地波动。为了让能量得到有效引导 通过复杂的交互网络，从而准确地激活必要的转变，通常 需要几百微秒，到毫秒，甚至到几十秒的动态。 几十年的生物物理学研究表明蛋白质可能具有特征能量 编码特定功能的景观。尽管理论和计算研究已经 极大地提高了我们对蛋白质能量景观的理解，现有的知识是 仍然非常有限。主导概念，例如构象选择模型和层次结构 能量景观（构象奴役）模型尚未得到充分理解 原子级。这主要是由于缺乏可靠的“预测”分子动力学采样 能够充分探索长期尺度蛋白质构象的技术 变化。 正交空间采样（OSS）方案，特别是其高阶泛化，允许 根据彻底采样增强的需要，系统地加速能量流。 初步研究表明，采样增强的数量级是合理的。 然而，OSS面临的主要挑战是缺乏严格的算法解决方案来确保 抽样稳健性。我们最近在自适应动态报告 (ADR) 方法方面的创新 发展揭示了这一挑战。在这个项目中，我们将系统地开发和 在蛋白质长期尺度的背景下改进这种新颖的“预测”采样策略 动力学并采用待开发的方法大规模定量探索蛋白质 构象动力学和破译蛋白质功能背后的生物物理原理 动态。 这项研究包括三个具体目标：（1）开发高阶正交空间回火 （HOOST）方法基于自适应动态报告（ADR）内核，以实现鲁棒性 生物分子长时尺度动力学的“预测”自由能采样； （二）理解 溶剂化波动在蛋白质动力学中的作用； (3) 理解其机制基础 人类葡萄糖激酶 (hGK) 调节。

Project Summary Protein aimlessly fluctuates in its surrounding. In order for energy to be effectively channeled through the complex interaction network and so accurately activate essential transitions, often hundreds of microseconds, to milliseconds, even to tens of seconds of dynamics are required. Several decades’ biophysical studies suggest that proteins likely possess characteristic energy landscapes that encode specific functions. Although theoretical and computational studies have greatly improved our understanding on protein energy landscape, the existing knowledge is still very limited. Dominant concepts, such as conformation selection model and hierarchical energy landscape (conformational slaving) model, have not been adequately understood at the atomistic level. This is largely due to lack of robust “predictive” molecular dynamics sampling technique that can enable adequate exploration of long-timescale protein conformational changes. The orthogonal space sampling (OSS) scheme, particularly its high order generalization, allows for systematic acceleration of energy flow as required for thorough sampling enhancement. Preliminary studies suggest that orders of magnitude of sampling enhancement are plausible. However a major challenge for OSS has been lack of rigorous algorithmic solution to ensure sampling robustness. Our recent innovation in the adaptive dynamic reporting (ADR) method development sheds light on this challenge. In this project, we will systematically develop and improve this novel “predictive” sampling strategy in the context of protein long-timescale dynamics and employ to-be-developed methods to quantitatively explore protein large-scale conformational dynamics and decipher biophysical principles underlying protein functional dynamics. This study includes three specific goals: (1) Developing high order orthogonal space tempering (HOOST) method based on the adaptive dynamic reporting (ADR) kernel to enable robust “predictive” free energy sampling of biomolecular long-timescale dynamics; (2) Understanding roles of solvation fluctuation in protein dynamics; (3) Understanding the mechanistic basis of human Glucokinase (hGK) regulation.