AF: Small: Motion Planning Techniques for Protein Motion

蛋白质运动在许多生化过程中起着至关重要的作用。例如，当蛋白质折叠到其天然的功能状态时，它们有时会经历影响其功能的关键构象变化，例如，疯牛病或阿尔茨海默病等疾病与蛋白质的错误折叠和聚集有关。当蛋白质在结合和非结合状态之间转换时，当它们与其他分子相互作用时，也会发生构象变化。了解这些运动过程的机制可能有助于洞察蛋白质如何以及为什么错误折叠，结合调节是如何通过蛋白质结构传达的，以及如何设计更有效的药物。例如，更好地了解蛋白质的错误折叠和聚集有可能为阿尔茨海默病、帕金森氏病、普里恩病等神经退行性疾病以及对社会产生重大影响的相关疾病提供洞察。该项目的一个重要且更直接的目标是在公开可用的蛋白质运动数据库中与社区共享新方法产生的详细结果。该项目将开发新的建模、模拟和分析工具，专门并应用一种新的计算方法来研究分子运动，该方法已经开发并在初步工作中与实验数据进行了验证。这种方法代表了分子动力学和蒙特卡罗模拟等提供详细的个体折叠轨迹的方法和提供全球景观统计的统计机械方法等技术之间的权衡。该方法源于机器人运动规划方法，构建了一个对许多(通常是数千条)运动路径进行编码的图。拟议的工作既包括进一步开发和优化技术的算法研究，也包括将这些技术应用于研究当前蛋白质科学感兴趣的问题所需的研究。虽然算法研究将由计算机科学家进行，但这些技术的应用和测试将受益于与目前正在研究这些问题的实验室的合作。主要研究目标包括：(I)开发新的和/或改进的构象、路径和路线图的度量和分析技术，可用于对更复杂的运动应用进行建模。这些方法将被验证并应用于蛋白质转换、诱饵数据库改进和配体结合。(2)模拟和模拟受限运动的新方法，并在合理需要的领域纳入更大的结合灵活性(目前的框架都不支持这两种方法)。这些方法将被应用于模拟蛋白质的转变和配体结合。(3)采用高性能计算以增加可研究系统的规模和复杂性的战略。

Protein motions play an essential role in many biochemical processes. For example, as proteins fold to their native, functional state, they sometimes undergo critical conformational changes that affect their functionality, e.g., diseases such as Mad Cow disease or Alzheimer's disease are associated with protein misfolding and aggregation. Proteins also undergo conformational change when interacting with other molecules as they transition between bound and unbound states. Knowledge of the mechanics of these motion processes may help provide insight into how and why proteins misfold, how binding regulation is communicated through the protein structure, and how to design more effective drugs. For example, a better understanding of protein misfolding and aggregation has the potential to provide insight into neurodegenerative diseases such as Alzheimer's disease, Parkinson's disease, prion diseases, and related diseases that have a major impact on society. An important, and more immediate, goal of the project is to share detailed results generated by the new methods with the community in a publicly available database of protein motions. This project will develop new modeling, simulation and analysis tools that specialize and apply a novel computational method for studying molecular motions that has been developed and validated against experimental data in preliminary work. This method represents a trade-off between methods such as molecular dynamics and Monte Carlo simulations that provide detailed individual folding trajectories and techniques such as statistical mechanical methods that provide global landscape statistics. The approach, derived from robotic motion planning methods, builds a graph that encodes many (typically thousands) of motion pathways. The proposed work involves both algorithmic research to further develop and optimize the techniques and research necessary to apply them to study issues of current interest in protein science. While the algorithmic research will be performed by computer scientists, the application and testing of the techniques will benefit from collaborations with labs currently studying these problems. The main research goals include: (i) The development of new and/or improved metrics and analysis techniques for conformations, pathways, and roadmaps that can be applied to modeling more complex motion applications. These methods will be validated and applied to protein transitions, decoy database improvement, and ligand binding. (ii) New methods for modeling and simulating constrained motion and incorporating greater bond flexibility in areas of legitimate need (neither supported by current framework). These methods will be applied to modeling protein transitions, and ligand binding. (iii) Strategies for employing high-performance computing to increase the size and complexity of the systems that can be studied.