Prediction of Electrophoretic Mobility of Analytes Using Abraham Solvation Parameters by Different Chemometric Methods

Abstract

Background: Quantitative structure-mobility relationships are proposed to estimate the electrophoretic mobility of diverse sets of analytes in capillary zone electrophoresis using Abraham solvation parameters of analytes, namely the excess molar refraction, polarizability, hydrogen bond acidity, basicity, and molar volume. Multiple linear regression (MLR) as a linear model, adaptive neuro-fuzzy inference system (ANFIS), and artificial neural network (ANN) methods were used to evaluate the non-linear behavior of the involved parameters. The applicability of the Abraham solvation parameters to the mobility prediction of analytes was studied employing various datasets consisting of organic acids, benzoate derivatives, pyridines, and ammoniums. Methods: To evaluate the simulation ability of the proposed models, datasets were subdivided into training and test sets in the ratio of 3:1. To evaluate the goodness of fit of the models, squared correlation coefficients (R-2) between experimental and calculated mobilities were calculated. Results: R-2 values were better than 0.78 for all datasets except for organic acids, in which the ANFIS model showed better ability to predict their mobility than that of MLR and ANN. In addition, the accuracy of the models is calculated using mean percentage deviation (MPD) and the overall MPD values for test sets were better than 15% for all models. Conclusion: The results showed the ability of the developed models to predict the electrophoretic mobility of analytes in capillary zone electrophoresis.