药物相互作用的网络药理学分析与预测

发布时间：2018-01-16 07:08

本文关键词：药物相互作用的网络药理学分析与预测　出处：《北京交通大学》2016年硕士论文　论文类型：学位论文

【摘要】：目前老龄化社会的到来以及多种慢性疾病及基础疾病的共存,大大促进了患者的联合用药,使得药物相互作用成为医疗中不可回避的实际问题。药物相互作用的理想状态是提高治疗效果,但实际上药物的联合使用常常会引起不良反应甚至严重副作用的发生。药物相互作用的影响因素很多,因此其作用机理是非常复杂的。尽管当前该方面的研究已有一些成果,但是人们对药物相互作用的机理仍处于模糊状态,该研究有待进一步探索。本文主要通过拓扑特性分析和分子机理分析发现药物相互作用的网络药理学特征,同时还提出了一种基于相似性的药物相互作用预测方法。主要工作如下：1.通过度分布、介数、集聚系数、紧密中心性和模块性等几个重要拓扑属性分析药物相互作用网络的拓扑特征,结果表明该网络是一个模块化的网络(模块度为0.523,平均集聚系数为0.226)。另外,药物相互作用网络中拓扑结构紧密的模块内的药物功能较相似,而模块与药物疾病的关系相对较弱。该研究提示药物相互作用在功能类似的药物之间形成了聚集的模块,可能存在相应的分子网络机理。2.从PPI网络、药物靶点、药物ATC分类和化学结构四个分子层面定性分析药物相互作用的分子网络机理,结果发现有相互作用的药物的靶点趋向于分布在同一分子网络拓扑模块内,且与药物对的ATC结构和化学结构相似性紧密相关。另外,本文还从这四方面分析不同类型的药物相互作用机理,发现药物相互作用的不同效应与其分子模块性、功能特性和化学结构相似性有直接的关系。3.该预测方法基于相似的药物有相似的相互作用的假设,从药物模块性和ATC结构的角度,通过相似性的测量和标准相互作用数据集的整合预测新的相互作用药物对。本预测方法效果较好,通过ROC曲线评价预测模型,得到其AUROC值分别为0.95和0.92。另外本方法还可以将预测得到的相互作用药物对追溯到标准集中对应的药物对,结合标准药物对的药理作用和临床效果分析预测得到的药物的相互作用效应。
[Abstract]:At present, the arrival of aging society and the coexistence of many chronic diseases and basic diseases greatly promote the combination of drug use. Drug interaction has become an unavoidable practical problem in medicine. The ideal state of drug interaction is to improve the therapeutic effect. But in fact, the combination of drugs can often lead to adverse reactions and even serious side effects. There are many factors affecting drug interaction. Therefore, the mechanism of drug interaction is very complicated. Although there have been some achievements in this field, the mechanism of drug interaction is still in a vague state. This study needs to be further explored. In this paper, the network pharmacological characteristics of drug interactions are found through topological analysis and molecular mechanism analysis. At the same time, a method of drug interaction prediction based on similarity is proposed. The main work is as follows: 1. Several important topological properties such as close-centrality and modularity are analyzed. The results show that the network is a modular network (modularity is 0.523). The average agglomeration coefficient was 0.226. In addition, the drug function in the modules with close topological structure in the drug interaction network was similar. The relationship between the module and drug diseases is relatively weak. This study suggests that drug interactions form aggregation modules between drugs with similar functions, and there may be corresponding molecular network mechanism .2. from the PPI network. The molecular network mechanism of drug interaction was qualitatively analyzed at four molecular levels: drug target, drug ATC classification and chemical structure. The results showed that the targets of the drugs with interaction tended to be distributed in the same molecular network topology module and were closely related to the similarity of the ATC structure and the chemical structure of the drug pair. In this paper, we also analyze different drug interaction mechanisms from these four aspects, and find out the different effects and molecular modularity of drug interaction. The prediction method is based on the hypothesis that similar drugs have similar interactions, from the perspective of drug modularity and ATC structure. Through the similarity measurement and the integration of the standard interaction data set to predict the new interaction drugs. This prediction method is effective, and the prediction model is evaluated by ROC curve. The AUROC values were 0. 95 and 0. 92 respectively. In addition, the predicted interaction pairs could be traced back to the corresponding drug pairs in the standard set. The pharmacological effect and clinical effect of standard drug were analyzed and predicted.
【学位授予单位】：北京交通大学
【学位级别】：硕士
【学位授予年份】：2016
【分类号】：R96

【相似文献】