QSPR with extended topochemical atom (ETA) indices, 3: Modeling of critical micelle concentration of cationic surfactants

The ability of surfactant molecules to facilitate solubilization of essential chemical entities accounts for their wide application in the pharmaceutical industry. The solubilization ability begins at concentrations exceeding the critical micelle concentration (CMC) which in turn is influenced by the molecular structure of the surfactants. Here, attempts have been made to explore the essential molecular fragments determining CMC of cationic surfactants using in silico technique. Different classes of descriptors (ETA and non-ETA) were correlated with the CMC data of a series of cationic surfactants to develop predictive models. Models were developed using genetic function approximation technique and were extensively validated employing different validation statistics. Comparative analysis was performed between the best model developed using the ETA descriptors and that obtained with the non-ETA variables. The results revealed that the external predictive potential of the ETA model was superior to the non-ETA model. The results also indicated that the ETA model could efficiently determine the CMC profile of the untested molecules. Finally, the two classes of descriptors were combined and the QSPR model thus obtained exhibited improved external predictive potential. The ETA descriptors implicated the importance of molecular size and their degree of unsaturation while the non-ETA descriptors signified impact of the bond order and the number of fragments bearing hydrogen atoms attached to carbon atoms with variable hybridization. Thus, addition of ETA descriptors improved the model quality extensively identifying essential molecular fragments contributing to their CMC profile and can be comprehensively utilized for property prediction of new cationic surfactants.

► The ETA indices have been used for modeling of logCMC values of 35 cationic surfactants.
► The ETA indices can be easily calculated from 2D representation of chemical structure.
► The external predictivity of the best ETA model is better than the corresponding non-ETA model.
► The use of ETA descriptors along with the non-ETA ones improves the statistical quality of the models.
► The models can be used for prediction of logCMC values of cationic surfactants.