PRAC - Ensembles of Molecular Dynamics Engines for Assessing Force Fields, Conformational Change, and Free Energies of Proteins and Nucleic Acids

利用Blue Waters Petascale资源，我们的研究重点是开发和应用准确的方法，使用Amber软件模拟生物分子。利用这些模拟的系综，我们可以有效地模拟生物分子的构象分布，以更好地了解它们的结构、动力学和相互作用，最终提供关于它们功能的详细信息。这些信息可以用来设计调节功能的药物，并帮助确定如何改变生物分子结构和动力学来影响功能。在过去的几年里，使用这些技术，我们已经展示了重复收敛各种RNA分子的构象分布(即在特定条件下采样的一组构象，例如在特定温度、pH或特定环境下)的能力。我们还改进了分子“力场”，允许对结构和动力学进行适当的建模。这通常可以让我们更好地了解生物分子机器，包括各种蛋白质和RNA分子是如何运作的。例如，我们的目标之一是了解核糖开关如何识别特定的代谢物，进而导致构象变化，从而改变这些代谢物的调节。另一个目标是更好地了解晶体中生物分子的结构和动力学，这将导致更好地理解结构和无序在蛋白质功能中的作用。开发的方法和“力场”被一个更大的社区广泛使用，旨在为生物分子结构、动力学和功能提供现实的见解。尽管目前并不是所有的研究小组都可以访问Blue Waters Petascale资源，但我们现在开发和应用的技术将成为更大的社区的例行公事，他们可以很容易地将这些方法应用于未来的资源。由于计算机能力和可获得性继续快速增长，更广泛的社区将在几年内看到这些技术的影响。在过去的几十年里，模拟方法的进步和计算能力的增加结合在一起，改变了我们对生物大分子的理解；我们可以观察小蛋白质折叠成正确的结构，我们可以帮助设计新的疗法，改进的模拟方法可以帮助提炼低分辨率或模糊的实验数据。我们应用的生物分子模拟方法通过暴露关于生物分子在许多大小和时间尺度上运动的不易测量的信息来增加实验数据，从最快的键振动到较慢的集体运动。这使得蛋白质和核酸作为时间函数的动态图景得以阐明，特别是在较小的尺度上。获得PB级计算资源使我们第一次能够充分探索复杂生物分子的结构、动态和能量景观。在我们目前的PRAC中，我们已经能够完全收敛DNA螺旋、RNA四核苷酸和四环的构象系综，我们评估和改进了力场，我们还开发了新颖的多维副本交换方法和分析工具。在经验丰富的琥珀开发商S团队的进一步合作下，我们的目标是破译蛋白质和核酸结构和功能的全貌，重点是核糖核酸、脱氧核糖核酸和蛋白质及其复合体。在Blue Waters上，我们将继续分层紧密地耦合高度GPU优化的分子动力学引擎，以全面绘制生物分子的构象、能量和化学景观。这不仅是为了评估、验证和改进目前可用的生物分子力场，而且还将为晶体和溶液中的DNA结构、RNA核糖开关的动力学和功能以及蛋白质-核酸相互作用提供新的见解。其他目标侧重于开发新的分析方法，并将模拟数据传播给更大的社区，以便进行更深入和更广泛的检查。

Using the Blue Waters petascale resource, our research centers on developing and applying accurate methods for the simulation of biomolecules using the AMBER software. Using ensembles of these simulations, we can efficiently model the conformational distributions of biomolecules to better understand their structure, dynamics and interactions, and ultimately this provides detailed information about their function. This information can be used to design drugs to modulate function and to help determine how to alter the biomolecular structure and dynamics to influence function. Over the past few years, using these technologies we have shown the ability to reproducibly converge the conformational distributions (i.e. the set of conformations sampled under a particular set of conditions, such as at a particular temperature, pH, or specific environment) of various RNA molecules. We have also improved the molecular "force fields" that allow proper modeling of the structure and dynamics. This generally allows us to better understand how biomolecular machines, including various protein and RNA molecules, function. For example, one of our goals is to understand how riboswitches recognize specific metabolites which in turn leads to conformational changes that alters regulation of those metabolites. Another goal is to better understand the structure and dynamics of biomolecules in the crystal and this will lead to a better understanding of the role of structure and disorder in protein function. The methods and "force fields" developed are in wide use by a larger community that is aiming to provide realistic insight into biomolecular structure, dynamics, and function. Although not all research groups have access to the Blue Waters petascale resource at this time, the technology we are developing and applying now will become routinely accessible to the larger community who can then easily apply these methods on future resources. Since computer power and accessibility continues to grow at a rapid pace, the broader community will see the impact of these technologies within a few years.Advances in simulation methods and increases in computational power have coupled together over the past few decades to transform our understanding of biological macromolecules; we can watch small proteins fold to their correct structure, we can help design new therapeutics, and improved simulation methods can help refine low resolution or ambiguous experimental data. The biomolecular simulation methods we apply add to experimental data by exposing information not readily measured about the motions of biomolecules across many size and time scales, ranging from the fastest bond vibrations to slower collective motions. This allows elucidation of the dynamic landscape of proteins and nucleic acids as a function of time, especially at smaller scales. Access to petascale computational resources allows us for the first time to fully explore the structural, dynamic and energetic landscape of complex biomolecules. In our current PRAC, we have been able to fully converge the conformational ensemble of DNA helices, RNA tetranucleotides and tetraloops, we assessed and improved the force fields, and we also developed novel multi-dimensional replica-exchange methods and analysis tools. With further collaboration of an experienced team of AMBER developer?s, we aim to decipher the full landscape of protein and nucleic acid structure and function, with a heavy focus on RNA, DNA and proteins and their complexes. On Blue Waters we will continue to hierarchically and tightly couple ensembles of highly GPU optimized molecular dynamics engines to fully map out the conformational, energetic and chemical landscape of biomolecules. This will be done not only to assess, validate and improve currently available biomolecular force fields, but also to provide novel insight into DNA structure in the crystal and in solution, RNA riboswitch dynamics and function, and also protein-nucleic acid interactions. Additional aims focus on the development of new analysis methods and dissemination of the simulation data to the larger community for deeper and broader inspection.