论文部分内容阅读
利用分子全息技术研究了129个5-羧基苯并咪唑类HCV NS5B聚合酶抑制剂的结构与活性之间的关系.讨论了分子碎片大小、碎片区分参数及全息长度对模型质量的影响.利用偏最小二乘法(partial least square,PLS)建立了一组以99个化合物为训练集的最优模型,该模型的交叉验证相关系数q~2=0.820,非交叉验证相关系数r~2=0.963,标准偏差SEE=0.213;用最优模型对由30个化合物组成的测试集进行预测,得到其相关系数r_(pred)~2=0.98,表明了该模型具有良好的预测能力及拟合能力.利用色码图对模型中不同原子及不同结构的贡献进行了解释,在此基础上根据最优HQSAR模型设计了几种具有良好抗HCV活性的苯并咪唑类HCV NS5B聚合酶抑制剂分子,为新型HCV NS5B聚合酶抑制剂的设计和优化提供了参考.
The relationship between the structure and activity of 129 5-carboxybenzimidazole HCV NS5B polymerase inhibitors was investigated by molecular holography. The effects of molecular fragment size, fragment differentiation parameters and holographic length on the model quality were discussed. A set of optimal models with 99 least squares (PLS) training sets were established by partial least squares (PLS). The correlation coefficient of this model was q ~ 2 = 0.820, r ~ 2 = 0.963, The standard deviation (SEE) was 0.213. The optimal model was used to predict the test set consisting of 30 compounds, and the correlation coefficient r_ (pred) ~ 2 = 0.98 was obtained, indicating that the model has good predictive ability and fitting ability. The color-coded diagrams explain the contributions of different atoms and different structures in the model. Based on this, we designed several benzimidazole HCV NS5B polymerase inhibitors with good anti-HCV activity according to the optimal HQSAR model. The design and optimization of HCV NS5B polymerase inhibitors provide a reference.