QSPR modeling of retention factor for some organic compounds in supercritical fluid chromatography

Supercritical fluid chromatography (SFC) combines the high diffusion coefficient of gas chromatography (GC) and the solubility properties of liquid chromatography. A very strong model, based on a quantitative structure-property relationship (QSPR) is developed using multiple linear regression approach and descriptors for determination of retention factor of some organic compounds. In the present study multiple linear regression (MLR) has been applied to modeling and prediction of retention factor of various compounds in SFC. In the first step, molecular descriptors were calculated and the redundant and irrelevant descriptors were omitted. Then the data set was divided into two groups, training and prediction sets. In the second step the stepwise multiple linear regression method was employed to screening and construction of a QSPR model. Finally, the cross validation method was used for the evalution of the generated models. We built four models for different percent of modifier (methanol). The obtained results of this work after cross-validation test are: 1) Q2=0.922, RMSE=0.130 2) Q2=0.917, RMSE=0.056 3) Q2=0.920, RMSE=0.014 4) Q2=0.931, RMSE=0.023 for 0%, 2%, 4% and 6% methanol, respectively.