Modeling The Capacity Factor of Analytes in Micellar Electrokinetic Chromatography Using Computational Descriptors

A multiple linear regression model is proposed to calculate capacity factor of analytes using structural features computed using HyperChem software. The chemical descriptors of analytes were computed using HyperChem software and regressed against the experimental capacity factors of analytes collected from the literature. The absolute average percentage deviation (AARD) and individual percentage deviation (IPD) were calculated as accuracy criteria. The accuracy of the proposed method was compared with that of a previously reported linear salvation energy relationship (LSER). The proposed method was tested on ten experimental data sets and mean standard deviation of AARDs were 48.5 20.4 and 130.1 79.7, respectively, for the proposed and LSER models in which the mean difference was statistically significant (p<0.01). The distribution of IPDs sorted in three subgroups, i.e. 45%, 45%-90%, and >90 %, shows the superiority of the proposed model over the LSER. A significant improvement in capacity factor modeling was achieved and the improvement factor is about 2.7. The descriptors could be easily computed and the calculations are straightforward. Therefore, the model is suggested to be employed in practice, however, the efforts should be continued until providing more accurate models.