典型抗结核病药物的QSAR研究

发布时间：2018-08-23 09:35

【摘要】：结核病是由结核杆菌感染引起的慢性传染病。解放后人民的生活水平不断提高，结核病已基本控制，但是近年来随着环境污染和艾滋病(HIV)的传播以及药物耐受，结核病发病率急剧增加。目前临床上使用治疗方案虽然可使85%以上的初治肺结核患者痊愈，但多存在不良反应、药物间相互作用、疗程长、对耐多药结核病无效和对潜伏态的结核杆菌作用不强等缺点。因此，需要深入了解目前抗结核药物的作用机制及耐药机制，以指导研发对持留菌和耐药菌更加有效的新型药物，实现对结核病的有效治疗与控制。新药的研发是一项周期长，耗资大，技术要求高，风险大的系统工程。从定量构效关系(QSAR)这一角度出发，首先从理论上预测其药理活性，再用于临床使用，有效地降低开发成本，并且能够有目的的设计合成新药。本文采用新型3D分子结构描述子——三维全息原子场作用矢量(3D-HoVAIF)，在不借助任何实验手段及无需样本构想重叠的前提下建立了结核病药物活性与其结构的定量关系，，取得了相当满意的结果。为新型抗结核病药物的研发提供了理论依据。本文的研究内容有以下几个方面：（1）采用3D-HoVAIF对25个香豆素-4乙酰苄肼类抗结核病药物分子进行结构参数化表征，同时结合逐步回归的方法对众多变量进行筛选，以得出相关性高的变量用于建立与其活性相关的QSAR模型。分别采用偏最小二乘回归(PLS)和多元线性回归(MLR)建模，为考虑模型稳定性同时对训练集和测试集进行双重验证。PLS建模和MLR建模的复相关系数(R_(cum))、留一法交互校验复相关系数(Q_(LOO))以及外部样本校验复相关系数(Q_(ext))分别为0.919、0.828、0.836和0.926、0.819、0.805。结果表明，3D-HoVAIF描述子能较好表征抗结核药物分子结构信息，为抗结核病药物的研发提供一定的理论基础。（2）采用3D-HoVAIF对115个酰肼类抗结核病药物分子进行结构参数化表征，同时结合逐步回归的方法对众多变量进行筛选，以得出相关性高的变量用于建立与其活性相关的QSAR模型。分别采用PLS和MLR建模，为考虑模型稳定性同时对训练集和测试集进行双重验证。PLS建模和MLR建模的三个相关系数R_(cum)~2、Q_(LOO)~2以及Q_(ext)~2分别为0.733、0.614、0.715和0.766、0.663、0.748。结果表明，3D-HoVAIF描述子能较好表征抗结核药物分子结构信息，所建QSAR模型具有良好稳定性和预测能力。（3）采用3D-HoVAIF对28个芳基酰胺类抗结核病药物分子进行结构参数化表征。该体系进行了三次训练集和测试集的划分，同时结合逐步回归的方法对众多变量进行筛选，以得出相关性高的变量用于建立与其活性相关的QSAR模型。三个体系均得到了良好的效果。（4）用3D-HoVAIF对另一芳基酰胺类抗结核病药物分子进行结构参数化表征，该体系有37个药物分子。该体系进行了两次次训练集和测试集的划分，同时结合逐步回归的方法对众多变量进行筛选，以得出相关性高的变量用于建立与其活性相关的QSAR模型。分别采用PLS和MLR建模，为考虑模型稳定性同时对训练集和测试集进行双重验证。结果表明，3D-HoVAIF描述子建立的模型取得了优于文献结果。该方法能较好地表征该类分子结构信息，有物化意义明确及结果易于解释等特点，值得进一步推广应用。
[Abstract]:Tuberculosis is a chronic infectious disease caused by the infection of Mycobacterium tuberculosis. After liberation, people's living standards have been improved and tuberculosis has been basically controlled. However, in recent years, with the spread of environmental pollution, HIV and drug tolerance, the incidence of tuberculosis has increased dramatically. TB patients recover, but there are many disadvantages, such as adverse reactions, drug interactions, long course of treatment, ineffective against MDR-TB and weak effect on latent TB. Therefore, it is necessary to understand the mechanism of action and drug resistance of anti-TB drugs in depth, so as to guide the development of new drugs which are more effective against retaining bacteria and drug-resistant bacteria. To achieve effective treatment and control of tuberculosis.
The research and development of new drugs is a systematic engineering with long period, high cost, high technical requirement and high risk. From the perspective of quantitative structure-activity relationship (QSAR), the pharmacological activity of new drugs is predicted theoretically, and then used in clinic. The new drugs can be designed and synthesized effectively and purposefully. The relationship between the activity and structure of tuberculosis drugs has been established without any experimental means and overlapping of sample concepts. The satisfactory results have been obtained, which provide a theoretical basis for the development of new anti-tuberculosis drugs.
The research contents of this paper are as follows:
(1) Twenty-five coumarin-4-acetylbenzylhydrazine antituberculosis drugs were characterized by 3D-HoVAIF, and many variables were screened by stepwise regression method to obtain high correlation variables for establishing QSAR model related to their activities. Partial least squares regression (PLS) and multiple linear regression (MLR) were used respectively. MLR modeling, in order to consider the stability of the model, both the training set and the test set are validated. The complex correlation coefficients (R_ (cum)) of PLS modeling and MLR modeling, the interactive check complex correlation coefficients (Q_ (LOO)) and the external sample check complex correlation coefficients (Q_ (ext)) are 0.919, 0.828, 0.836 and 0.926, 0.819, 0.805, respectively. The narrator can better characterize the molecular structure information of anti-tuberculosis drugs and provide a theoretical basis for the research and development of anti-tuberculosis drugs.
(2) 115 hydrazide anti-tuberculosis drugs were characterized by 3D-HoVAIF, and many variables were selected by stepwise regression method to obtain high correlation variables for establishing QSAR model related to their activities. PLS and MLR were used to model the QSAR model respectively, and the training set and measurement were carried out simultaneously to consider the stability of the model. The three correlation coefficients R_ (cum) ~2, Q_ (LOO) ~2 and Q_ (ext) ~2 of PLS and MLR were 0.733, 0.614, 0.715 and 0.766, 0.663 and 0.748, respectively. The results showed that the 3D-HoVAIF descriptor could well characterize the molecular structure information of anti-tuberculosis drugs, and the QSAR model had good stability and predictive ability.
(3) Twenty-eight arylamide antituberculosis drugs were characterized by 3D-HoVAIF. The system was divided into three training sets and test sets, and many variables were selected by stepwise regression method to obtain high correlation variables for establishing QSAR model related to their activities. Good results have been achieved.
(4) Structural parameterization of another aryl amide antituberculosis drug molecule was carried out by 3D-HoVAIF. There were 37 drug molecules in the system. The system was divided into two training sets and test sets. At the same time, many variables were screened by stepwise regression method to obtain high correlation variables for the establishment of its active phase. The results show that the model established by 3D-HoVAIF descriptor is superior to the results of literature. The method can well characterize the molecular structure information of this kind, and has the characteristics of clear physical and chemical meaning and easy interpretation of the results. Further popularization and application.
【学位授予单位】：陕西科技大学
【学位级别】：硕士
【学位授予年份】：2013
【分类号】：R52;R91

【参考文献】