Improving the Accuracy of the Amber Force Field for Biomolecular Simulation

斯托尼布鲁克大学（SBU）的卡洛斯西默林得到了化学系化学理论、模型和计算方法项目的一个奖项的支持，以开发改进的蛋白质分子计算模型。蛋白质是生物学中的重要组成部分。它们在人类和所有其他生物体中执行20，000多种不同的化学和机械功能。理解生物学详细机制的一个主要挑战是蛋白质不是刚性物体。它们的运动和扭曲方式对其功能至关重要。这通常是分子之间发生关键识别的地方，因为分子相互适应并相互拉动，以确认它们与正确的伴侣相互作用，就像握手时双手适应一样。这些关键的运动可以通过原子级详细的计算机模拟来研究，使用经典近似的基本量子力学力，称为力场。 力场使蛋白质及其相互作用的精确模拟成为可能，否则即使在最大的超级计算机上也难以进行研究。除了基础生物学研究，精确的力场对于药物发现和新型纳米级和仿生材料的设计至关重要。 Simmerling教授与布鲁克海文国家实验室的Qin Wu合作，正在对生物物理和生物分子系统的分子动力学模拟中使用的最重要模型之一琥珀力场进行全面，系统，基于物理的重塑。该项目的目标是显着提高力场的准确性和保真度，同时保持其长期模拟大型生物分子系统的能力。新参数化的力场将免费供模拟社区下载和使用，用于在XSEDE超级计算机和小型实验室集群上运行的最广泛使用的分子动力学模拟代码中。该项目的外联工作包括通过SBU全纳教育中心协调，定期前往HBCU机构进行研究和招聘，该项目的目标是系统地和一致地解决现有蛋白质经典力场的几个关键限制，以便在不显着增加计算量的情况下提高其准确性。复杂性目前的模型在某些情况下有效，但在许多其他情况下与实验不一致，包括氨基酸特定性质的定量再现，如螺旋倾向。Simmerling教授与布鲁克海文国家实验室的Qin Wu合作，正在开发更准确的蛋白质骨架结构和动力学能量分布描述，具有更好的序列依赖结构和动力学，以扩展和重新参数化琥珀力场。该项目有三个主要目标：（1）通过在溶液中使用高级QM计算拟合多维扫描来改进骨架能量学的描述，（2）扩展Amber力场参数库以包括生物学中遇到的非标准氨基酸和那些经常用作实验探针的氨基酸，（3）开始解决在处理短程货车范德华相互作用方面长期存在的弱点，利用量子力学能量分解方法的最新进展来训练替代的函数形式。这些应该提高模型再现侧链旋转异构体对主链构象偏好的已知影响的能力。新的力场模型正在通过广泛使用的Amber分子动力学（MD）模拟程序进行验证，记录和分发，并且可以独立下载以与其他广泛使用的MD代码结合使用。 教育和推广的重点是招聘和提供实验室研究经验的学生从代表性不足的群体在SBU，并鼓励学生在科学和研究的兴趣，通过外展旅行HBCU机构。

Carlos Simmerling of Stony Brook University (SBU) is supported by an award from the Chemical Theory, Models and Computational Methods program in the Chemistry Division to develop improved computational models of protein molecules. Proteins are the workhorses of biology. They perform more than 20,000 different chemical and mechanical functions in humans, and in all other living organisms. A major challenge in understanding the detailed mechanisms of biology is that proteins are not rigid objects. They move and twist in ways that are essential for their function. This is often where the crucial recognition between molecules takes place, as molecules adapt and pull on each other to confirm that they are interacting with the correct partner, much as two hands adapt during a handshake. These crucial motions can be studied via atomically-detailed computer simulations, using classical approximations to the underlying quantum mechanical forces, known as force fields. Force fields enable the accurate simulation of proteins and their interactions, studies that would otherwise be intractable, even on the largest supercomputers. In addition to fundamental biological studies, accurate force fields are essential for drug discovery and the design of novel nanoscale and biomimetic materials. Professor Simmerling, in collaboration with Qin Wu of Brookhaven National Laboratory, is performing a comprehensive, systematic, physics-based reshaping of one of the most important models used in molecular dynamics simulations of biophysical and biomolecular systems, the Amber force field. The goal of the project is to significantly improve the force field's accuracy and fidelity, while preserving its ability to simulate large biomolecular systems for long times. The newly-parameterized force field will be made freely available for download and use by the simulation community, for use within the most widely-used molecular dynamics simulation codes running on XSEDE supercomputers and small lab clusters. Outreach efforts for the project include regular research and recruitment trips to HBCU institutions, coordinated through the SBU Center for Inclusive Education, and the recruitment of undergraduates from underrepresented minorities to participate in the research.The goal of this project is to systematically and consistently address several key limitations of existing protein classical force fields in order to improve their accuracy without dramatically increasing computational complexity. Current models work in some cases, but fail to agree with experiment in many others, including quantitative reproduction of amino-acid specific properties such as helical propensity. Professor Simmerling, in conjunction with collaborator Qin Wu of Brookhaven National Laboratory, is developing more accurate descriptions of the energy profiles for protein backbone structure and dynamics, with better sequence-dependent structure and dynamics, in order to extend and reparameterize the Amber force field. The project has three principal aims: (1) to improve the description of backbone energetics through fitting to multidimensional scans using high-level QM calculations in solution, (2) to expand the Amber force field parameter library to include non-standard amino acids encountered in biology and those frequently used as experimental probes, and (3) to begin to address longstanding weaknesses in the treatment of short-range van der Waals interactions, by taking advantage of recent advances in quantum mechanical energy decomposition methods to train alternate functional forms. These should improve the ability of the model to reproduce the known influence of side chain rotamer on backbone conformational preference. The new force field models are being validated, documented, and distributed via the widely-used Amber molecular dynamics (MD) simulation program, and can be independently downloaded for use in conjunction with other widely-used MD codes. Education and outreach is focused on recruiting and providing lab research experiences for students from underrepresented groups at SBU, and encouraging student interest in science and research through outreach trips to HBCU institutions.