论文部分内容阅读
为了研究热力学性质的构效关系,选取了60种白藜芦醇类似物的热力学性质作为分析对象,引入了主量子拓扑指数{~0P_2,~0P_4,~0P_3’}、{∑~0P}。通过对比分析,兼顾模型的简洁性,确定{∑~0P}为模型的基本自变量。结合正弦级数的特点,引入{sin(k∑~0P)}(k=1,2,3…)到模型中期望消除残差。以调整判定系数最大(R_(adj)~2)为目标,在控制变量多重共线性并兼顾变量最少的条件下,建立了各热力学性质构效关系模型。结果表明:引入sin(5∑~0P)、sin(8∑~0P)到模型中,能较好地消除残差,提高模型的预测能力,所有建立的模型的相关系数均大于0.99,甚至达到1。采用交叉验证方法对模型的稳健性进行了检验,并对样本外分子的热力学性质进行了预测,结果表明所建模型具有良好的稳健性和预测能力。
In order to study the structure-activity relationship of thermodynamic properties, the thermodynamic properties of 60 resveratrol analogues were selected as the analytic objects, and the main quantum topological indices {~ 0P_2, ~ 0P_4, ~ 0P_3 ’} and {Σ ~ 0P} were introduced. Through comparative analysis, taking into account the simplicity of the model, to determine {Σ ~ 0P} as the basic argument of the model. Combining with the characteristics of the sine series, {sin (kΣ ~ 0P)} (k = 1,2,3) is introduced into the model to eliminate the residuals. In order to adjust the maximum coefficient of judgment (R_ (adj) ~ 2), under the condition that the control variables are multicollinearity with the least amount of variables, the model of the relationship between the thermodynamic properties and the structure-activity relationship is established. The results show that the introduction of sin (5Σ ~ 0P) and sin (8Σ ~ 0P) into the model can eliminate the residuals and improve the predictive ability of the model. The correlation coefficients of all the models are greater than 0.99, 1. The robustness of the model was tested by cross-validation, and the thermodynamic properties of the samples were also predicted. The results show that the proposed model has good robustness and predictability.