有机物毒性与其特征结构关联性研究

发布时间：2018-11-12 12:20

【摘要】：构效关系方法广泛应用于药物化学和环境化学研究中,在药物先导物设计和污染物的环境风险评价,生态评价以及人体健康评价等领域发挥重要作用。构效关系研究是在化合物的结构与性质/活性之间建立定量描述数学关系式从而实现性质/活性参数的预测,该方法是弥补化合物性质参数实验数据短缺的重要途径,同时能够有效降低实验数据测定中的耗时,费力和活体生物消耗。本工作提出了系列距离矩阵,变形欧式矩阵和性质矩阵,利用范数算法进行了多角度数据挖掘,得到了系列范数描述符,再利用分布构效关系方法建立模型研究139个化合物(饱和一元醇和酮)对老鼠的急性口服毒性和143种组蛋白脱乙酰酶抑制剂的生物活性(抗癌活性)。同时使用留一/五折/十折交叉验证、外部验证、Y随机验证和平均绝对误差验证等方式进行模型评价。针对饱和一元醇和酮对老鼠的急性口服毒性预测模型1研究结果表明,模型预测精度较高,相关系数为0.882,留一/五折/十折交叉验证的相关系数分别是0.833、0.832和0.838,训练集、测试集的相关系数分别为0.879和0.866。统计结果和相关验证结果表明,本工作建立的有机物毒性预测模型1准确稳定。针对组蛋白脱乙酰酶抑制剂的生物活性预测模型2研究结果表明,活性预测值与实验值符合良好,留一/5折/10折交叉验证的相关系数与原始模型非常相近,相关统计参数和应用域的验证结果均表明本方法对于组蛋白脱乙酰酶抑制剂的活性预测具有较高的准确性和稳定性同时,本工作利用杠杆法对两个模型的应用域分别进行了考察,结果表明两个模型均有较宽的应用域,有可能推广应用。统计参数和文献对比结果表明本工作提出的新方法能够有效预测有机物的毒性以及组蛋白脱乙酰酶抑制剂的生物活性。最后,根据提出的构效关系方法,本工作初步设计了两个活性较高的组蛋白脱乙酰酶抑制剂。综上,本工作提出的范数描述符可以有效表达分子结构和生物活性/毒性之间的关系,基于范数描述符建立的模型可以有效预测化合物的毒性和HDACs抑制活性,而且该方法也可以用于化合物的环境生态危险性评价,对于抗癌先导药物的合成也有重要指导作用。
[Abstract]:Structure-activity relationship (STR) is widely used in drug chemistry and environmental chemistry. It plays an important role in drug precursor design, environmental risk assessment of pollutants, ecological evaluation and human health evaluation. The study of structure-activity relationship is to establish a quantitative description mathematical relation between the structure and property / activity of compounds, so as to predict the properties / activity parameters. This method is an important way to make up for the shortage of experimental data on the properties of compounds. At the same time, it can effectively reduce the time-consuming, labor and living organism consumption in the measurement of experimental data. In this paper, the series distance matrix, deformed Euclidean matrix and property matrix are proposed. The multi-angle data mining is carried out by norm algorithm, and the series norm descriptors are obtained. The acute oral toxicity of 139 compounds (saturated monools and ketones) and the bioactivity (anticancer activity) of 143 histone deacetylase inhibitors (histone deacetylase inhibitors) in mice were studied. At the same time, the model is evaluated by one / 50% discount / ten fold cross validation, external validation, Y random validation and average absolute error verification. The results of acute oral toxicity prediction model 1 for mice with saturated monools and ketones showed that the prediction accuracy of the model was high, and the correlation coefficient was 0.882. The correlation coefficients of one / 50% discount / ten fold cross-validation are 0.833 / 0.832 and 0.838 respectively. The correlation coefficients of training set and test set are 0.879 and 0.866 respectively. The statistical results and related verification results show that the model 1 is accurate and stable. The bioactivity prediction model 2 for histone deacetylase inhibitors showed that the predicted activity values were in good agreement with the experimental values, and the correlation coefficients of a 50% discount / 10 fold cross-validation were very close to those of the original model. The statistical parameters and the validation results of the application domain show that this method has high accuracy and stability in predicting the activity of histone deacetylase inhibitors. In this paper, the application fields of the two models are investigated by using the lever method. The results show that the two models have a wide range of application fields, and it is possible to extend the application. The results of statistical parameters and literature comparison show that the proposed new method can effectively predict the toxicity of organics and the bioactivity of histone deacetylase inhibitors. Finally, two histone deacetylase inhibitors with high activity were designed according to the proposed structure-activity relationship method. In conclusion, the proposed norm descriptor can effectively express the relationship between molecular structure and bioactivity / toxicity, and the model based on norm descriptor can effectively predict the toxicity of compounds and the inhibitory activity of HDACs. The method can also be used to assess the environmental and ecological risk of compounds, and also has an important guiding role in the synthesis of anticancer lead drugs.
【学位授予单位】：天津科技大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：TQ460.1;X13

【相似文献】