Element:Software:Enabling Millisecond-Scale Biomolecular Dynamics

几十年来，基于分子动力学（MD）的计算机模拟方法一直被用于在分子水平上理解化学和生物化学现象。MD是一个非常强大的工具，它使科学家能够了解对生命至关重要的分子的行为，如蛋白质和核酸。医学也被用于了解疾病和开发新药。然而，MD在可以研究的系统的大小和可以模拟的时间上都是有限的。许多与生命相关的复杂现象涉及的系统太大，无法用MD来研究，或者需要对系统进行更长时间的跟踪。作为传统分子动力学的替代方法，不连续分子动力学（DMD）已被证明在研究生物分子过程方面更为有效。然而，到目前为止，由于DMD无法利用现代并行计算机的优势，它的使用受到了限制。该项目将开发下一代并行DMD软件，使研究涉及更大系统和更长的时间尺度的复杂分子现象成为可能。对这些过程的详细了解将大大促进新材料和药物的发展，并促进人类健康。项目团队结合了计算和实验专业知识，成功地开发和验证了一个健壮的并行DMD软件框架。软件和结果将与计算模拟社区以及整个科学和工程界积极共享，为国家网络基础设施生态系统的能力、容量和凝聚力做出贡献。此外，该项目的成果将用于旨在教育和包容传统上在高等STEM教育中代表性不足的少数民族的推广工作。该项目旨在开发一个开放的软件框架，通过实现并行不连续分子动力学（DMD）包，实现多肽和多肽模拟物的多毫秒动态模拟。不像目前的分子动力学（MD），其特点是有限的模拟时间尺度，不连续分子动力学（DMD）假设相互作用事件之间的粒子的弹道运动，并能够在更长的时间尺度上研究现象。为了演示该方法，该项目将(1)开发现有串行DMD代码的并行版本，以便将模拟时间从数百微秒延长到几毫秒；(2)扩展和完善了现有的DMD肽力场，增加了非天然肽和类肽的参数；(3)开发将传统MD相互作用势转换为DMD的软件。项目团队拥有该项目所需的互补专业知识，包括复杂聚合物和类肽的粗粒度模型和力场，蛋白质自组装和肽-蛋白质结合过程的MD模拟，蛋白质结合肽和类肽的合成，以及热力学和动力学结合参数的测量。这项研究产生的工具将使科学和工程界能够模拟和研究非常长时间尺度的现象，如生物聚合物折叠、聚集和聚集抑制、纤维形成和蛋白质结合。这个工具箱显示了巨大的希望，不仅加速了生物材料计算设计的创新，而且影响了专注于高度复杂系统的分子模拟社区，直到细胞水平的动力学。值得注意的是，该项目是美国国家科学基金会持续科学创新网络基础设施（CSSI）的理想选择，因为它(i)通过提供用户友好的开源计算工具，有助于提高国家网络基础设施生态系统的能力、容量和凝聚力，（ii）积极吸引CI专家和我们工具箱的测试人员，他们可能是它的最终用户。（iii）提高我们目前在开发生物活性肽和类肽方面的能力，（iv）制定计划和指标，鼓励测量进展并认证共享工具和结果的质量，以及（iv）制定战略，将广泛获取与长期社区驱动的发展和进步相结合。该奖项反映了美国国家科学基金会的法定使命，并通过使用基金会的知识价值和更广泛的影响审查标准进行评估，被认为值得支持。

Computer simulation methods based on Molecular Dynamics (MD) have been used for decades to understand chemical and biochemical phenomena at the molecular level. MD is a very powerful tool that has enabled scientists to understand the behavior of molecules crucial for life such as proteins and nucleic acids. MD has also been used to understand diseases and develop new drugs. However, MD is limited in both the size of the systems that can be studied and the amount of time that can be simulated. Many complex phenomena relevant to life involve systems too large to study with MD, or require following the system for much longer times. An alternative to traditional MD called discontinuous molecular dynamics (DMD) has been shown to be much more efficient to study biomolecular processes. To date, however, the use of DMD has been limited due to its inability to take advantage of modern parallel computers. This project will develop a next-generation parallel DMD software that will enable the study of complex molecular phenomena involving larger systems and longer time scales. Detailed knowledge of such processes will considerably advance the development of new materials and drugs, and human health. The project team combines the computational and experimental expertise to successfully develop and validate a robust parallel DMD software framework. The software and results will be actively shared both with the computational simulation community and with the scientific and engineering community at large, contributing to the capability, capacity, and cohesiveness of the national cyber-infrastructure ecosystem. Furthermore, the results of this project will be used in outreach efforts geared toward the education and inclusion of minorities traditionally underrepresented in higher STEM education.This project aims to develop an open software framework that enables multi-millisecond dynamic simulations of peptides and peptide-mimetics by implementing a parallel discontinuous molecular dynamics (DMD) package. Unlike current molecular dynamics (MD), which features limited simulation timescales, discontinuous molecular dynamics (DMD) assumes ballistic motion of the particles between interaction events and enables the study of phenomena across much longer time scales. To demonstrate the approach, the project will (1) develop a parallel version of existing serial DMD codes to enable extending simulation times from hundreds of microseconds to several milliseconds; (2) extend and improve the available DMD peptide force field, adding parameters for non-natural peptides and peptoids; and (3) develop software for translating interaction potentials from traditional MD to DMD. The project team possesses the complementary expertise necessary for this project, including coarse-grained models and force fields for complex polymers and peptoids, MD simulation of protein self-assembly and peptide-protein binding processes, synthesis of protein-binding peptides and peptoids, and measurement of thermodynamic and kinetic binding parameters. The tools resulting from this research will allow the scientific and engineering community to model and study very long time-scale phenomena, such as biopolymer folding, aggregation and inhibition of aggregation, fibril formation, and protein-binding. This toolbox shows great promise to not only accelerate innovation in the computational design of biomaterials, but also to impact the molecular simulation community focusing on highly complex systems, up to cell-level dynamics. Notably, this project is ideal for the National Science Foundation's Cyber-infrastructure for Sustained Scientific Innovation (CSSI), as it (i) contributes to the capability, capacity, and cohesiveness of the national cyberinfrastructure ecosystem by providing user-friendly open-source computational tools, (ii) actively engages CI experts and testers of our toolbox, who would potentially be its ultimate users, (iii) advances our current capabilities in developing bioactive peptides and peptoids, (iv) establishes plans and metrics that encourage measurement of progress and certify the quality of shared tools and results, and (iv) devise strategies to combine wide-access with long-term community-driven development and progress.This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.