感染丙型肝炎病毒(Hepatitis C virus,HCV)可引起慢性肝炎、肝硬化、甚至肝细胞癌.开发新型、高效的抗HCV药物迫在眉睫.半最大效应浓度EC50是衡量药物活性的重要参数.在本研究中,基于结构—活性关系建立了121种硫脲衍生物抗HCV活性的分...感染丙型肝炎病毒(Hepatitis C virus,HCV)可引起慢性肝炎、肝硬化、甚至肝细胞癌.开发新型、高效的抗HCV药物迫在眉睫.半最大效应浓度EC50是衡量药物活性的重要参数.在本研究中,基于结构—活性关系建立了121种硫脲衍生物抗HCV活性的分类模型.硫脲衍生物分成训练集(61种衍生物)和测试集(60种衍生物).从每种硫脲衍生物计算4885个分子参数之后,采用二元Logistic回归的Wald向前逐步方法得到包含5个分子参数的分类模型.所建分类模型对训练集整体预测精度(precision)为93.4%,对测试集整体预测精度91.7%.本文所得分类模型与文献报道模型相比,具有更好的预测硫脲衍生物抗HCV活性的能力.展开更多
The reactivity parameters,Q and e,in the Q-e scheme reflect the reactivities of a monomer(or a radical)in free-radical copolymerizations.By applying multiple linear regression(MLR)analysis,the optimal quantitative str...The reactivity parameters,Q and e,in the Q-e scheme reflect the reactivities of a monomer(or a radical)in free-radical copolymerizations.By applying multiple linear regression(MLR)analysis,the optimal quantitative structure-activity relationship(QSAR)model for the reactivity parameter lnQ was developed based on five descriptors(NAF,NOF,EαLUMO,EβHOMO,and EβLUMO)and 69 monomers with the root mean square(rms)error of 0.61.The optimal MLR model of the parameter e obtained from five descriptors(TOcl,NpN,NSO,EαHOMO and DH)and 68 monomers produced rms error of 0.42.Compared with previous models,the two optimal MLR models in this paper show satisfactory statistical characteristics.The feasibility of combining 2D descriptors obtained from the monomers and 3D descriptors calculated from the radical structures(formed from monomers+H )to predict parameters Q and e has been demonstrated.展开更多
文摘感染丙型肝炎病毒(Hepatitis C virus,HCV)可引起慢性肝炎、肝硬化、甚至肝细胞癌.开发新型、高效的抗HCV药物迫在眉睫.半最大效应浓度EC50是衡量药物活性的重要参数.在本研究中,基于结构—活性关系建立了121种硫脲衍生物抗HCV活性的分类模型.硫脲衍生物分成训练集(61种衍生物)和测试集(60种衍生物).从每种硫脲衍生物计算4885个分子参数之后,采用二元Logistic回归的Wald向前逐步方法得到包含5个分子参数的分类模型.所建分类模型对训练集整体预测精度(precision)为93.4%,对测试集整体预测精度91.7%.本文所得分类模型与文献报道模型相比,具有更好的预测硫脲衍生物抗HCV活性的能力.
基金supported by the National Natural Science Foundation of China(No.21472040)the Scientific Research Fund of Hunan Education Department(Nos.16A047 and 18A344)the Open Project Program of Hunan Provincial Key Laboratory of Environmental Catalysis & Waste Regeneration(Hunan Institute of Engineering)(2018KF11)
文摘The reactivity parameters,Q and e,in the Q-e scheme reflect the reactivities of a monomer(or a radical)in free-radical copolymerizations.By applying multiple linear regression(MLR)analysis,the optimal quantitative structure-activity relationship(QSAR)model for the reactivity parameter lnQ was developed based on five descriptors(NAF,NOF,EαLUMO,EβHOMO,and EβLUMO)and 69 monomers with the root mean square(rms)error of 0.61.The optimal MLR model of the parameter e obtained from five descriptors(TOcl,NpN,NSO,EαHOMO and DH)and 68 monomers produced rms error of 0.42.Compared with previous models,the two optimal MLR models in this paper show satisfactory statistical characteristics.The feasibility of combining 2D descriptors obtained from the monomers and 3D descriptors calculated from the radical structures(formed from monomers+H )to predict parameters Q and e has been demonstrated.