论文部分内容阅读
为提高脂肪醇化合物对梨形四膜虫急性毒性的预测精度,提出基于定量结构-活性关系(QSAR)原理的脂肪醇化合物对梨形四膜虫急性毒性预测方法。运用遗传算法筛选出5种分子描述符作为变量,采用多元线性回归方法和最小二乘-支持向量机方法建立基于该5种分子描述符的脂肪醇化合物对梨形四膜虫急性毒性的预测模型。对所建立的模型进行内部验证和外部验证,两种模型的复相关系数、留一法交互验证系数分别为0.984、0.979和0.985、0.982,对外部预测样本的复相关系数和外部测试集交互验证系数分别为0.978、0.977和0.979、0.979。结果表明,所建QSAR模型均具有较好的稳健性、预测能力和泛化性能。LS-SVM模型在精度上略优于ML-R模型,而MLR模型更为简单和方便。
In order to improve the prediction accuracy of fatty alcohol compounds against Tetrahymena acute toxicity, a method for the prediction of acute toxicity of fatty alcohol compounds to Tetrahymena tettigoniridae based on the quantitative structure-activity relationship (QSAR) was proposed. Five kinds of molecular descriptors were screened by genetic algorithm (GA) as variables, and the prediction model of acute toxicity of tetrazaprine was established based on the five kinds of molecular descriptors using multivariate linear regression method and least square support vector machine . The internal model verification and external verification of the established model, the two models of the complex correlation coefficient, leaving a law of mutual verification coefficient of 0.984,0.979 and 0.985,0.982, respectively, the external correlation coefficient of the sample and the external test set interactive verification The coefficients were 0.978, 0.977 and 0.979, 0.979, respectively. The results show that the proposed QSAR model has good robustness, predictive ability and generalization performance. LS-SVM model is slightly superior to ML-R model in accuracy, while MLR model is more simple and convenient.