GNEIMO: Generalized Internal Coordinate Molecular Dynamics Methods

描述(申请人提供)：分子动力学(MD)模拟是结构生物学中的一种强大的计算工具，广泛用于了解蛋白质的构象变化和多肽的折叠。然而，使用笛卡尔动力学模型进行分子动力学模拟受到大蛋白质总模拟时间尺度为几十纳秒的限制。另一方面，生物过程需要微秒的模拟时间。内坐标分子动力学(ICMD)算法已经被开发为能够实现更大的模拟时间步长，它们在长时间尺度的模拟中显示出巨大的前景。尽管ICMD技术前景看好，但由于内部坐标模型的额外数学复杂性，ICMD技术进展甚微。我们提出用ICMD算法来解决关键的瓶颈问题。我们建议开发和验证只允许冻结键长、键角和键长的广义NEIMO(GNEIMO)ICMD方法，并使用ICMD方法使用模型粗化策略进行更广泛的构象搜索。我们还提出了ICMD算法在如下应用中的性能特征：1)保持天然蛋白质结构，2)改进接近天然同源的结构模型，3)折叠α螺旋和β发夹多肽，4)小蛋白质的构象变化。总而言之，该项目内的各种研究将为有效利用ICMD提供算法和路线图。我们将为这些算法与广泛使用的MD软件包NAMD的集成奠定基础，以便更广泛地传播。

DESCRIPTION (provided by applicant): Molecular dynamics (MD) simulations is a powerful computational tool in structural biology, widely used for understanding conformational changes in proteins, and folding of peptides. However MD simulations using Cartesian dynamics model is limited by the total simulation time scale being in tens of nanoseconds for large proteins. Biological processes on the other hand need microseconds of simulation time. Internal Coordinate Molecular Dynamics (ICMD) algorithms have been developed to enable larger simulation time-steps and they show great promise in long time scale simulations. Despite their promise, ICMD techniques have made little progress due in large part to the additional mathematical complexity of internal coordinate models. We propose to address and solve the key bottleneck problems with ICMD algorithms. We propose to develop, validate Generalized NEIMO (GNEIMO) ICMD methods that allow freezing of only bond lengths, bond angle and bond lengths and use of ICMD methods for wider conformational search using model coarsening strategies. We also propose to characterize the performance of ICMD algorithms for the applications such as 1) maintaining the native protein structure, 2) refinement of a near native homology structural models, 3) folding of alpha helical and beta hairpin peptides and 4) conformations changes in small proteins. Together, the various studies within this project will provide algorithms and a roadmap for the effective use of ICMD. We will lay the basis for integration of these algorithms with the widely used MD software package NAMD for wider dissemination.