多种虚拟筛选工具的比较研究及其组合运用

发布时间：2018-05-28 11:30

本文选题：虚拟筛选 + 分布式计算　；参考：《天津科技大学》2016年硕士论文

【摘要】：随着计算技术的飞速发展和世界各地药物的需要急剧增加,药物研发已在过去60年里成为一个热点问题。计算机辅助药物设计(CADD)在药物发现过程中起到了日益重要的作用。传统的药物筛选方法对于研究机构和制药公司来说难以负担。所以,在计算机上的虚拟筛选经常用于优化高通量筛选实验。这可以使药物研发的成本压低到一较低的水平。分布式计算是指利用分布式系统来解决计算问题,其中的基本组件通信通过在分布式计算机传递消息来协调它们的行动。任务被分成许多子任务,其中的每一个由一个或多个处理器解决。对于每个分布式作业有一个管理主机负责拆分工作,提交子任务和整合结果。依赖于程序,管理节点可以在本地主机上或在远程主机上运行。硬件和软件调试是在建立高通量虚拟筛选通道过程中重要的一环,其中计算节点的兼容性,数据同步,合并文件和简便的脚本都是重要的因素。适当地选择打分函数也是我们方法验证的过程中遇到的问题。我们选择了已知很多抑制剂的Aurora-A激酶和HDM-1内酰胺酶等作为靶点来比较分析各种各样的筛选方法如Ligandfit,GLIDE HTVS,GLIDE SP,GLIDE XP和Surflex,来说明基于分布式计算的虚拟筛选方法的合理性。Aurora-A,也被称为丝氨酸/苏氨酸-蛋白激酶6,是与细胞增殖和癌症发展的调控相关的Aurora激酶家族的一个成员。HDM-1内酰胺酶是是耐药细菌针对内酰胺类抗生素分泌的一类酶,使β—内酰胺环裂解而被破坏,失去抗菌活性。本文利用具有660万分子的类药分子库,针对NDM-1受体进行虚拟筛选。体系的筛选速度和能力再次得到了肯定,得到了 13类421个分子。经过目视检查,选择其中12个进行购买与合成,并进行了生物活性测试与结构优化改造,获得了有较高抑制活性的候选化合物,通过对其进一步的优化与改造,阐述了衍生物结构与活性之间的关系,以上研究将为药物研发提供新方法和新思路。
[Abstract]:With the rapid development of computing technology and the rapid increase of drug demand all over the world, drug research and development has become a hot issue in the past 60 years. Computer aided drug design (CAD) plays an increasingly important role in drug discovery. Traditional drug screening methods are unaffordable for research institutions and pharmaceutical companies. Therefore, virtual filtering on computers is often used to optimize high-throughput screening experiments. This can reduce the cost of drug development to a lower level. Distributed computing is the use of distributed systems to solve computing problems, in which the basic components of communication through the transmission of messages in the distributed computer to coordinate their actions. Tasks are divided into subtasks, each of which is resolved by one or more processors. For each distributed job, a managed host is responsible for splitting, submitting subtasks and consolidating results. Depending on the program, the management node can run on a local host or on a remote host. Hardware and software debugging is an important part of the establishment of high-throughput virtual filtering channels, in which computing node compatibility, data synchronization, merging files and simple scripts are all important factors. Proper selection of scoring functions is also a problem in the process of our method verification. We selected Aurora-A kinase and HDM-1 lactamases of many known inhibitors as targets to compare and analyze various screening methods such as Ligandfitn GLIDE, GLIDE SPSPIDE, GLIDE XP and Surflex. to illustrate the reasonableness of the virtual screening method based on distributed computing. Aurora-A. Also known as serine / threonine protein kinase 6, a member of the Aurora kinase family associated with cell proliferation and cancer development, HDM-1 lactamases are a class of enzymes secreted by drug-resistant bacteria against lactam antibiotics. The 尾-lactam ring was destroyed by cleavage and lost antibacterial activity. In this paper, a drug-like library with 6.6 million molecules was used for virtual screening of NDM-1 receptors. The screening speed and ability of the system were confirmed again, and 421 molecules of 13 classes were obtained. After visual inspection, 12 of them were selected for purchase and synthesis, and the bioactivity test and structural optimization were carried out. Candidate compounds with high inhibitory activity were obtained, which were further optimized and modified. The relationship between the structure and activity of derivatives is expounded. The above research will provide new methods and new ideas for drug research and development.
【学位授予单位】：天津科技大学
【学位级别】：硕士
【学位授予年份】：2016
【分类号】：R91;TP391.7

【参考文献】