海量药物数据的靶标关系预测研究

发布时间：2018-06-26 13:00

本文选题：网络药理学 + 虚拟筛选　；参考：《华东理工大学》2014年硕士论文

【摘要】：网络药理学为药物研发带来了新的理论指导,其思想是构建一个信息全面的生物信息网络,采用计算机技术对网络进行模拟分析,多方位考察药物在网络中的作用,降低毒副作用,提高治疗效果,从而提高新药研发的成功率,降低研发成本。虚拟筛选采用计算机技术模拟先导化合物的筛选过程,从海量药物数据库中匹配出推荐的先导化合物。目前应用广泛的几种虚拟筛选技术都是通过三维化学空间来分析,而很多靶标蛋白的三维结构都还没有测量出来,无法得以应用。另外,目前国际上的一些主流药物数据信息库共享信息有限,无法构建信息全面的药理网络。为解决这些问题,本文首先对网络药理学研究现状进行分析,设计了一个包含国际主流药物数据库信息以及实验信息实时关联入库的信息全面的药物生物化学信息库。本文进而分析了药物靶标关系预测研究现状,遵循网络药理学的理论指导,构建基于药物生物化学信息库数据的药理网络,提出两种混合相似度预测方法,在构建的药理网络中全面考虑药物的作用,取得了更好的药物靶标关系预测效果。最后,本文对药物生物信息库数据进行筛选,对两种混合方法进行验证分析,在处理海量药物数据时采用Hadoop分布式平台以提高运算效率。
[Abstract]:Network pharmacology has brought new theoretical guidance for drug research and development. Its idea is to construct a comprehensive information biological information network, to simulate and analyze the network by computer technology, and to investigate the role of drugs in the network. Reduce toxic side effects, improve therapeutic effect, thus improve the success rate of new drug research and development, reduce R & D costs. Virtual screening simulates the screening process of lead compounds and matches the recommended lead compounds from massive drug databases. At present, several virtual screening techniques are widely used to analyze them by three-dimensional chemical space. However, many of the three-dimensional structures of target proteins have not been measured and can not be applied. In addition, some mainstream drug data databases share limited information, so it is impossible to build a comprehensive pharmacological network. In order to solve these problems, this paper first analyzes the current situation of online pharmacology research, and designs a comprehensive drug biochemistry information base which includes the international mainstream drug database information and the real-time correlation of experimental information into the database. In this paper, the current research situation of drug target relationship prediction is analyzed, and the pharmacological network based on the data of drug biochemistry information base is constructed according to the theoretical guidance of network pharmacology, and two hybrid similarity prediction methods are proposed. The effect of drug was considered in the pharmacological network, and a better prediction effect of drug target relationship was obtained. Finally, the data of drug bioinformatics database are screened, the two mixed methods are verified and analyzed, and Hadoop distributed platform is adopted to improve the computational efficiency when dealing with massive drug data.
【学位授予单位】：华东理工大学
【学位级别】：硕士
【学位授予年份】：2014
【分类号】：R96-39

【参考文献】