QSPR study for estimation of acidity constants of some aromatic acids derivatives using multiple linear regression (MLR) analysis

A very simple, strong, descriptive and interpretable model, based on a quantitative structure-property relationship (QSPR), is developed using multiple linear regression approach and quantum chemical descriptors derived from AM1-based calculations (MOPAC7.0) for determination of the acidity constants of some aromatic acid derivatives. By molecular modeling and calculation of descriptors, three significant descriptors related to the pKa values of the acids, were identified. These are related to the partial charges at each atom in Oδ−-Hδ+ bond (pchgHδ+ and pchgOδ− −) and the changing of bond length in O-H molecular structures. A multiple linear regression (MLR) model based on 74 molecules as a training set has been developed for the prediction of the acidity constants of some aromatic acids using these quantum chemical descriptors. The effects of these theoretical descriptors on the acidity constants are discussed. The pKa values of aromatic acids generally decreased with increasing positive partial charges of acidic hydrogen atom. A model with low prediction error and high correlation coefficient was obtained. This model was used for the prediction of the pKa values of some aromatic acids (33 test acids) which were not used in the modeling procedure. The model obtained demonstrates excellent fit statistics and gives accurate predictions. The average relative error (RE¯%) of prediction set is lower than 1% and square correlation coefficient (R2) is 0.9882.