Algorithmic improvements in large scale polarizable QM/MM simulations

大规模极化 QM/MM 模拟的算法改进

基本信息

批准号：
10547634
负责人：
Xintian Feng
金额：
$ 62.32万
依托单位：
Q-CHEM, INC.
依托单位国家：
美国
项目类别：
财政年份：
2019
资助国家：
美国
起止时间：
2019-01-01 至 2024-07-31
项目状态：
已结题

来源：
https://reporter.nih.gov/project-details/10547634
关键词：
Algorithmic Software Algorithms Biochemical Reaction Biological Catalysis Chemical Models Chemicals Chemistry Code Complex Computational algorithm Computer Assisted Computer Models Computer software Data Development Drug Design Electrostatics Environment Evaluation Free Energy Freedom Geometry Goals Grain Hybrids Industry Intuition Literature Methodology Methods Modeling Molecular Nature Nobel Prize Nuclear Outcome Periodicity Pharmaceutical Preparations Phase Physiological Polymers Potential Energy Privatization Process Proteins Quantum Mechanics Reaction Research Sampling Solvents Spottings Surface System algorithm development base chemical reaction computer code computerized tools cost design drug development drug discovery improved intermolecular interaction molecular dynamics molecular mechanics molecular modeling novel practical application quantum simulation tool

项目摘要

Project Summary Modeling of chemical reactivity in heterogeneous environments such as protein pockets and complex solvents is an essential part of a drug discovery workﬂow. However, such modeling is challenging, due to large system sizes and necessity of extensive sampling of environment degrees of freedom. The goal of this project is to develop a suite of efﬁcient, accurate and scalable computational tools based on the polarizable quantum me- chanics / effective fragment potential (QM/EFP) methodology that will provide academic and private industry users with fast and robust software for the computational characterization of free energy proﬁles of chemical reactions in complex condensed phase systems. Phase II of this project builds upon the outcomes of a success- ful completion of Phase I, in which the team has developed algorithms and computer codes that dramatically decrease the computational cost of EFP and QM/EFP simulations by employing fast multipole method (FMM). In Phase II the team will further improve the efﬁciency of FMM-QM/EFP codes by implementing robust par- allel algorithms. Modeling of chemical transformations will be enabled by development of analytic nuclear gradients and second derivatives. Additionally, FMM-QM/EFP will be interfaced with polarizable continuum models (PCM) and extended to periodic boundary conditions that will provide users with complimentary tools for modeling long-range electrostatic and polarization interactions. New methodology will be validated on established and emerging data for mechanisms and energetics of solution-phase and enzymatic reactions.

项目概要对蛋白质袋和复杂溶剂等异质环境中的化学反应进行建模是药物发现工作流程的重要组成部分然而，由于系统庞大，这种建模具有挑战性。对环境自由度进行广泛采样的规模和必要性该项目的目标是开发一套基于可极化量子机制的高效、准确和可扩展的计算工具机械学/有效碎片潜力（QM/EFP）方法将为学术和私营行业提供用户可以使用快速、强大的软件来计算表征化学的自由能分布该项目的第二阶段建立在成功的成果之上。第一阶段的全面完成，该团队开发了算法和计算机代码，显着通过采用快速多极子方法 (FMM) 降低 EFP 和 QM/EFP 模拟的计算成本。在第二阶段，团队将通过实施稳健的标准来进一步提高 FMM-QM/EFP 代码的效率等位基因算法将通过分析核的发展来实现。此外，FMM-QM/EFP 将与可极化连续谱连接。模型（PCM）并扩展到周期性边界条件，这将为用户提供免费工具用于模拟长程静电和极化相互作用的新方法将得到验证。溶液相和酶反应的机制和能量学的已建立和新数据。