CAREER: Toward Boltzmann-Weighted Protein Ensembles Using Novel Computations

该CAREER项目由生物科学理事会分子与细胞生物科学部分子生物物理学和数学与物理科学理事会化学部理论与计算化学项目共同支持，其目标是继续开发和使用新的计算方法来平衡生物大分子，特别是蛋白质的采样。最终目标是生成典型采样的蛋白质结构的集合，表现出完全的波动。虽然人们普遍认为蛋白质结构的完整采样是一个极具挑战性的问题，但最近的技术进步表明这一目标最终可能实现。最近由Zuckerman及其同事开发的Resolution Exchange算法及其变体代表了一类特别有前途的算法，它们采用粗粒度模型来大大加快对蛋白质标准原子细节表示的严格统计抽样（基于玻尔兹曼因子）。粗粒度蛋白质模型的使用是至关重要的，因为这些模型可以被证明允许极其快速的配置空间的完整采样，需要不到一周的单处理器模拟少于100个残基的蛋白质。除决议交换方法外，还将采用新的收敛性评估和重新加权方法。此外，高温采样可以直接纳入分辨率交换模拟。教育计划是继续为生物物理学专业的学生写一篇关于统计力学的介绍性文章，目标是初级研究生或高级本科生。虽然统计力学可以说是分子生物物理学和模拟的基础科学，但对于来自各种背景的学生（即跨学科研究生课程的在校生）都能访问的文本，以及保持学生掌握基本概念所需的严谨性和普遍性，仍然是一个突出的需求。因此，本文将侧重于与生物物理学相关的分子现象，避免了许多研究生统计力学文本中对自旋系统和相变的强调。此外，文本也将主要依赖于基本的概率概念，而不是更抽象的热力学术语，作为传统的。概率论的概念是广泛适用的，使学生能够很好地掌握从熵和自由能到动力学和平衡之间的联系的概念。在整个文本中，与传统命名法的对应关系将被明确，因为这对于科学交流和文献理解是绝对必要的。

The objective of this CAREER project, jointly supported by Molecular Biophysics in the Division of Molecular and Cellular Biosciences in the Directorate for Biological Sciences and the Theoretical and Computational Chemistry Program in the Division of Chemistry in the Mathematical and Physical Sciences Directorate, is to continue development and use of novel computational approaches for equilibrium sampling of biomacromoleces, especially proteins. The final goal is to generate ensembles of canonically sampled protein structures, exhibiting full fluctuations. While it is widely appreciated that the full sampling of protein configurations is an extremely challenging problem, recent technical advances suggest the goal may finally be reachable. The Resolution Exchange algorithm and its variants, recently developed by Zuckerman and co-workers, represent a particularly promising class of algorithms for employing coarse-grained models to greatly accelerate rigorous statistical sampling (based on the Boltzmann factor) of standard atomically detailed representations of proteins. The use of coarse-grained protein models is critical, because these can be shown to permit extremely rapid full sampling of configuration space requiring less than one week of single-processor simulation for proteins with fewer than 100 residues. In addition to the Resolution Exchange approach, novel convergence assessment and re-weighting approaches will be employed. High-temperature sampling, moreover, can be directly incorporated into Resolution Exchange simulations.The education plan is to continue writing an introductory text on statistical mechanics for students of biophysics aimed at beginning graduate students or advanced undergraduates. Although statistical mechanics is arguably the fundamental science of molecular biophysics and simulation, there is an outstanding need for a text which is accessible to students from a wide variety of backgrounds (i.e., enrollees in inter-disciplinary graduate programs) and which also maintains the rigor and generality needed for students to grasp the fundamental concepts. The text will therefore focus on the molecular phenomena pertinent to biophysics, avoiding the emphasis on spin systems and phase transitions in many graduate statistical mechanics texts. Further, the text also will rely primarily on basic probability concepts rather than the more abstract terminology of thermodynamics, as has been traditional. The probability concepts are broadly applicable, permitting the student a strong grasp of concepts ranging from entropy and free energy to the connection between dynamics and equilibrium. Throughout the text, the correspondence with traditional nomenclature will be made clear, since this is absolutely necessary for scientific communication and the understanding of the literature.