论文部分内容阅读
Quantum chemistry parameters of 28 alkyl(1-phenylsulfonyl) cycloalkane-carboxy- lates were computed at the 6-31G* level in fully optimal manner using B3LYP method of density functional theory (DFT). With GQSARF2.0 program, the correlation equations that can predict n-octanol/water partition coefficient (lgKow) were developed using the structural and thermodynamic parameters of 28 alkyl(1-phenylsulfonyl) cycloalkane-carboxylates with experimental data of lgKow as theoretical descriptors; the correlation coefficient (R2) was 0.9452 and the cross-validation squared correlation coefficient (Rcv2) 0.9312. Furthermore, a four-variable model from MEDV was obtained, of which R2 = 0.9497 and Rcv2 = 0.9388. The models were validated by variance inflation factor (VIF) and t-test. Cross-validation indicates that the correlation and predicting ability of the model based on both DFT method and MEDV are more advantageous than those obtained from semi-empirical AM1 method.
Quantum chemistry parameters of 28 alkyl (1-phenylsulfonyl) cycloalkane-carboxy-lates were computed at the 6-31G * level in fully optimal manner using B3LYP method of density functional theory (DFT). With GQSARF2.0 program, the correlation equations that can predict n-octanol / water partition coefficient (lgKow) were developed using the structural and thermodynamic parameters of 28 alkyl (1-phenylsulfonyl) cycloalkane-carboxylates with experimental data of lgKow as theoretical descriptors; the correlation coefficient (R2) was 0.9452 and the Cross-validation squared correlation coefficient (Rcv2) 0.9312. The models were validated by variance inflation factor (VIF) and t-test. Cross -validation indicates that the correlation and predictability ability of the model based on both DFT method and MEDV are more advantageous than those obtained from semi-empirical AM1 method.