Institute of Forensic Medicine, Faculty of Medicine, University of Niš, Niš, Serbia.
Institute of Molecular Genetics and Genetic Engineering, University of Belgrade, Belgrade, Serbia.
Talanta. 2018 Feb 1;178:656-662. doi: 10.1016/j.talanta.2017.09.064. Epub 2017 Sep 23.
A method for the prediction of retention indices of pesticides using the Monte Carlo method and with optimal molecular descriptors based on local graph invariants and the SMILES notation of studied compounds has been presented. Quite satisfactory results were obtained with the proposed method, since a robust model with good statistical quality was developed. The predictive potential of the applied approach was tested and the robustness of the model was proven with different methods. The best calculated QSPR model had following statistical parameters: r = 0.9182 for the training set and r = 0.8939 for the test set. Structural indicators defined as molecular fragments responsible for the increases and decreases of gas chromatographic retention indices activity were calculated.
本文提出了一种使用蒙特卡罗法和基于局部图不变量和研究化合物的 SMILES 符号的最佳分子描述符预测农药保留指数的方法。该方法得到了相当令人满意的结果，因为开发了一个具有良好统计质量的稳健模型。应用方法的预测能力进行了测试，并通过不同的方法证明了模型的稳健性。最佳计算的 QSPR 模型具有以下统计参数：训练集的 r = 0.9182，测试集的 r = 0.8939。计算了定义为导致气相色谱保留指数活性增加和减少的分子片段的结构指标。
