构建分子动力学模拟研究平台分析PGRN与TNFR的分子识别机理

发布时间：2018-05-02 10:56

本文选题：生物大分子相互作用 + PGRN与TNFR的结合模式　；参考：《山东大学》2012年硕士论文

【摘要】：研究机体中各种生物大分子的相互作用方式,是理解生命活动基本机制的基础,以调控生物学活性的受体为靶位的肽类或模拟肽类配基药物研发是生物学和医学中最领先的研究领域之一。分子动力学模拟作为一种重要的重要工具,可以模拟生物大分子体系的运动行为、构象改变机制以及研究分子及其配体、分子各组成部分之间的相互作用,解释和阐明生物大分子的序列信息以及微观尺度上复杂的、动态的生物过程,以传统实验数据为基础,在更深层次上解决相关的科学问题,弥补传统实验技术的不足,目前已广泛应用于蛋白质、核酸等生物大分子微观尺度的结构与功能研究之中。生物大分子之间的识别/结合/催化等过程一般都是发生在微秒时间尺度甚至更长时间,而分子动力学模拟需要利用飞秒时间步长来模拟计算才可精确刻画分子运动过程,消耗的计算资源是普通个人计算机与工作站所无法提供的,必须利用大型集群的众多处理器进行并行计算,来实现几万原子至百万原子病毒体系的数亿步计算模拟任务,完成分析关键蛋白质分子的动力学行为,细胞膜表面配体受体间的结合过程及其机理以及病毒衣壳蛋白质体系的组装动力学过程等。然而,在这些实际应用首先要面临的一个问题就是如何合理的完成计算模拟任务在并行集群上的部署优化,构建完整的分子动力模拟研究平台,充分利用大型集群计算资源,高效地完成研究工作。本文首先基于MOSIX2并行操作系统构建了计算能力为千亿次PC集群,利用NAMD软件进行了万原子体系皮秒时间尺度的分子动力学模拟部署测试；随后利用了山东大学(山东省)高性能计算中心的浪潮TS10000十万亿次集群,完成了万原子体系纳秒时间尺度的部署测试；最后在国家超级计算济南中心使用了神威4000A百万亿次集群以及“神威蓝光”千万亿次超级计算机,使用NAMD和GROMACS软件完成了万原子体系近微秒级以及百万原子纳秒级的超大规模的部署测试,并完成了同时利用近十万处理器核心的副本交换分子动力学模拟测试分析。从整个部署测试结果来看,普通小型集群适于做分子动力学模拟任务的一些准备工作,如模型的构建、参数的优化等,而十万亿次集群上适合运行小体系的计算模拟以及大体系的计算模拟准备工作,对超大体系则必须要部署到百万亿次及千万亿次集群上,利用大规模的计算资源在可接受的时间范围内完成模拟任务。分子动力学模拟所得数据是在分子中每一个原子在三维空间中随时间的运动轨迹,所以只有三维图像才能有效地表示这些大分子所包含的信息,而以往专业的可视化技术与设备都是极其昂贵的,所以本论文基于廉价的3D VISION图形解决方案,搭建了生物大分子3D显示与分析平台,用于分析自身免疫性疾病相关蛋白质分子识别的动态机理。基于已有的实验发现,生长因子(PGRN)可以结合于肿瘤坏死因子受体(TNFR),从而抑制TNF-a介导的致炎效应,本论文通过动力学模拟确定了PGRN与TNFR2的结合模式与决定PGRN与TNFR2结合的关键氨基酸。通过构建PGRN虚拟突变体,计算模拟进一步证实PGRN结合TNFR2并发挥抗炎功能对这些氨基酸的依赖性和静电作用介导的PGRN/TNFR2结合模式。本研究深入探讨PGRN结合TNFR2根本原因,加深了对PGRN/TNFR2的分子识别过程的认识,同时也为进一步研究PGRN调控TNFR2信号途径的分子机制和Atsttrin等PGRN来源的新靶位药物设计提供结构学方面的参考与依据,促进以TNFR2为靶位、以PGRN为模板抗炎药物的临床应用转化,增强我国原创性重组蛋白药物的开发能力。
[Abstract]:The study of the interaction of various biological macromolecules in the body is the basis for understanding the basic mechanism of life activities. The research and development of peptides and analogue ligand drugs targeting the receptors of biological activities is one of the most leading research fields in biology and medicine. Molecular dynamic simulation as an important and important tool can be used as an important tool. In order to simulate the motion behavior of the biological macromolecular system, the mechanism of conformation change and the interaction between the molecules and their ligands and the components of the molecules, the sequence information of the biological macromolecules and the complex and dynamic biological processes on the micro scale are explained and explained on the basis of the traditional experimental data, and the correlation is solved on the deeper level. The scientific problems, which make up for the deficiency of traditional experimental technology, are now widely used in the study of the structure and function of the microscale of biological macromolecules, such as protein, nucleic acid and other biological macromolecules.
The process of recognition / binding / catalysis between biological macromolecules usually occurs at microsecond time scales or even longer, and molecular dynamics simulation needs to simulate calculation by using femtosecond time step length to accurately depict the molecular motion process. The computational resources consumed are not provided by ordinary human computers and workstations. Using a large number of processors in a large cluster to carry out parallel computing to achieve hundreds of millions of simulation tasks of the tens of thousands of atom to millions of atomic viruses, the dynamic behavior of the key protein molecules, the binding process between the ligand receptors on the cell membrane surface and its mechanism, and the assembly kinetics of the virus capsid protein system are completed. However, one of the first problems to be faced with in these practical applications is how to optimize the deployment of computational simulation tasks on a parallel cluster, build a complete platform for research on molecular dynamics simulation, make full use of the computing resources of large clusters, and efficiently complete the research work.
This paper first builds a computing power of hundreds of billions of PC clusters based on the MOSIX2 parallel operating system, and uses NAMD software to carry out the molecular dynamics simulation deployment test of the picosecond time scale of the ten thousand atomic system, and then uses the wave of the high performance computing center of the Shandong University (Shandong) to complete the ten million atomic system, which has completed the ten thousand atomic system. The deployment test of nanosecond time scale; finally, using the Shen Wei 4000A million billion times cluster and the "Shen Wei blue light" supercomputer in the National Supercomputing Ji'nan center, using NAMD and GROMACS software to complete the large scale deployment test of the near microsecond level of the ten thousand atom system and the million atom nanosecond level. At the same time, using the replica exchange molecular dynamics simulation test analysis at the core of nearly one hundred thousand processors, the general small cluster is suitable for the task of molecular dynamics simulation from the result of the whole deployment test.
Some preparations, such as the construction of the model, the optimization of the parameters, and the computing simulation for running small systems in the one hundred thousand billion cluster and the computing simulation preparation for the large system, must be deployed to the million and the millions of billions of clusters on the super large system, and the computing resources of the large rules can be completed within the acceptable time range. Simulation task.
The data obtained by molecular dynamics simulation are the trajectories of each atom in the three-dimensional space with time, so only three dimensional images can effectively express the information contained by these large molecules, and the previous professional visualization technology and equipment are extremely expensive, so this paper is based on the cheap 3D VISION graphical solution. A 3D display and analysis platform for biological macromolecules has been set up to analyze the dynamic mechanism of molecular recognition of proteins related to autoimmune diseases.
Based on the previous experiments, the growth factor (PGRN) can bind to the tumor necrosis factor receptor (TNFR) and inhibit the inflammatory effect mediated by TNF-a. In this paper, the binding mode of PGRN and TNFR2 and the key amino acids that determine the binding of PGRN to TNFR2 are determined by dynamic simulation. The simulation is further proved by the construction of a PGRN virtual mutant. Real PGRN combined with TNFR2 and exerts the PGRN/TNFR2 binding mode mediated by anti-inflammatory function to these amino acids and electrostatic action. This study deepened the understanding of the fundamental cause of PGRN binding to TNFR2, deepened the understanding of the molecular recognition process of PGRN/TNFR2, and further studied the molecular mechanism and Atsttrin of the PGRN regulation of TNFR2 signal pathways. The new target drug design, such as PGRN, provides the reference and basis for structural studies, and promotes the transformation of TNFR2 as the target, the clinical application of PGRN as a template anti-inflammatory drug, and the enhancement of the development ability of the original recombinant protein in our country.

【学位授予单位】：山东大学
【学位级别】：硕士
【学位授予年份】：2012
【分类号】：R373

【相似文献】