Computer modeling of the rate of glycine conjugation of some benzoic acid derivatives: a QSAR study

The rate of the glycine conjugation of 10 para- and 8 meta-monosubstituted, and 5 disubstituted benzoic acids in rat liver and kidney mitochondria was modeled with the aid of computer-assisted methods. A total of 75 numerical descriptors was chosen and calculated for each compound and multiple linear regression analysis was used to generate the equations that relate the structural features to the biological activities. The best model for the prediction of the log M-liv, contains five descriptors and its statistics are n=23, r=0.979, SE=0.14 and F=79 Also, five descriptors appeared in the model of the log M-kid and its statistics are n=23, r=0.982, SE=0.15 and F=72. Among different descriptors appearing in both models the partial charges of the most negative atom (PCHNEG) showed the largest mean effect on the glycine conjugation.