Advanced Electrostatic Computation in Molecular Dynamics

描述(申请人提供)：计算硬件和分子建模技术的进步彻底改变了我们模拟静电相互作用的能力，静电相互作用在生物分子的构象稳定性、结构、折叠和功能中发挥着重要作用。吸引理论和商业兴趣的这些发展的一个重要的近期应用是药物设计，在药物设计中，成功的对接需要形状和静电互补。目前，基于Poisson-Boltzmann方程(PBE)的连续介质静电描述提供了模拟保真度和计算成本的最佳组合，然而，与高带电系统的非线性行为和介电界面处的解收敛相关的数值问题已经影响了精度和计算时间。因此，基于PBE的求解器在能量最小化、蒙特卡罗和分子动力学程序中的应用受到了限制。在第一阶段，利用边界协调网格成功地解决了静电解的数值稳定性，特别是表面的梯度贡献，从而实现了可靠收敛和准确的力预测。高负荷系统的可靠收敛的挑战也得到了解决。这些进展是在自适应笛卡尔网格结构上实施的，该结构在多重网格实施、网格生成和解适应方面提供了与竞争网格安排(即网格和非结构化四面体网格)相比的独特内在优势。第二阶段的工作建立在这些成功的基础上，提供了侧重于药物设计和包装的额外能力，以促进向医疗和制药行业的最终用户过渡和分发软件工具。预计支持这一目标的主要技术发展是：(I)将制定和实施方法，以更准确和/或更快地计算静电相互作用或结合势和能量，从而促进药物筛选和设计工作中更高的可靠性和吞吐量。(Ii)将短程作用力纳入分子柔性模型，从而为药物设计、应用和了解生物功能提供更完整的分子动力学描述。(3)将开发新的评估指标的方法，以评估对接概率和拒绝诱饵，以及估计这些指标的梯度/灵敏度的有效公式。在药物设计的背景下，这些梯度将指示有利于选择性结合的药物几何形状和电荷分布的变化。我们的生物分子应用将建立在我们目前最先进的PBE解算器的优势之上，为高电荷和大规模系统(如核酸及其组件，如核糖体和核糖体)提供准确和快速的静电溶剂化自由能、结合自由能、表面静电势和导出的定量指标的预测。 与公共健康相关：研究工作将开发和提供软件工具以及分析和可视化技术，这些技术：(I)针对改进的蛋白质和药物设计而量身定做，(Ii)可用于将溶剂化生物分子的生物功能与其几何、结构和静电特性联系起来。对于药物应用，该分析将为设计者提供提高药物亲和力和特异性所需的信息，确保其与靶点结合并拒绝诱骗位置，从而最终降低药物开发成本和时间，并在减少副作用的情况下提高药物疗效。更深入地了解生物分子的物理化学和几何性质及其环境与生物功能之间的关系，有助于增强生物信息学工具，在分子水平上更好地从基础上理解疾病的发展和对抗疾病的手段。

DESCRIPTION (provided by applicant): Advances in computational hardware and molecular modeling techniques have revolutionized our ability to simulate electrostatic interactions, which play a fundamental role in the conformational stability, structure, folding and function of biomolecules. One important near-term application of these developments drawing both theoretical and commercial interests is drug design where successful docking requires both shape and electrostatic complementarity. Currently, the continuum electrostatic description based upon the Poisson- Boltzmann Equation (PBE) offers the best combination of modeling fidelity and computational cost, however, numerical issues associated with nonlinear behavior in highly charged systems and solution convergence at the dielectric interface, have impaired accuracy and calculation time. As a result, adoption of PBE-based solvers in energy minimization, Monte Carlo and molecular dynamics codes has been limited. In Phase I, the numerical stability of the electrostatic solution, particularly the gradient contributions from the surface, were successfully addressed using a boundary-conforming mesh so that reliably convergent and accurate force predictions are achieved. The challenge of reliable convergence for highly charged systems was also resolved. These advances were implemented on an adaptive Cartesian mesh structure that offers unique intrinsic advantages over competing grid arrangements (i.e. lattices and unstructured tetrahedral grids) with regard to multigrid implementation, mesh generation and solution adaptation. The Phase II effort builds upon these successes by providing additional capabilities focused on drug design and packaged to facilitate transition and distribution of the software tools to end-users in the medical and pharmaceutical industries. The main technical developments envisioned to support this goal are: (i) Methods will be formulated and implemented to calculate the electrostatic interaction or binding potential and energy with greater accuracy and/or speed, thus promoting higher reliability and throughput in drug screening and design efforts. (ii) Short- range forces will be incorporated to model molecular flexibility thus providing a more complete description of the molecular dynamics for drug design application and understanding of biologicial function. (iii) New methods for evaluating metrics to assess docking probability and reject decoys will be developed along with an efficient formulation for estimating the gradients/sensitivities of these metrics. In the context of drug design these gradients would indicate changes to the drug geometry and charge distribution favorable for selective binding. Our biomolecular applications will build upon strengths of our current state-of-the-art PBE solver in providing accurate and fast predictions of electrostatic solvation free energies, binding free energies, surface electrostatic potential and derived quantitative metrics for highly charged and large-scale systems such as nucleic acids and its assemblies such as nucleosome and ribosome. PUBLIC HEALTH RELEVANCE: The research effort will develop and provide software tools and analysis and visualization techniques that: (i) are tailored towards improved protein and drug design and (ii) can be used to relate biological function of solvated biomolecules to its geometric, structural and electrostatic properties. For drug applications, the analysis will provide designers with the information needed to enhance drug affinity and specificity, ensuring it binds to target sites and rejects decoy locations, thus, ultimately, lowering drug development costs and times, and improving drug efficacy with reduced side-effects. Improved insights into the relationship between the physiochemical and geometric properties of biomolecules and their environment to biological function are useful for enhancing bioinformatics tools and achieving a better foundational understanding of the progression of diseases at the molecular level and the means to counter them.