In-silico prediction of gas chromatographic retention indices of some terpenols

A quantitative structure-retention relationship study based on multiple linear regression technique was carried out to investigate the gas chromatographic retention indices (RIs) of some terpenols on the HP 5 ms fused silica column. A collection of 75 terpene alcohols was chosen as dataset. The data were divided into two groups; a training set and a prediction set consist of 60 and 15 molecules, respectively. The best-selected descriptors that appear in the models are; the Randic index order 1, Kier shape index order 2, total charge weighted partial negatively charged surface area, and fractional atomic charge weighted partial positive surface area. These descriptors can encode different features of molecules that are responsible for their steric, electronic, and lipophilicity interactions. The best-obtained model had statistics of R(2)(t) = 0.959 and R(2)(p) = 0.952. The reliability of the model was evaluated by using the leave-many-out cross-validation method (Q(2) = 0.957 and SPRESS = 46.427) as well as by y-scrambling and jackknife test. Furthermore, the chemical applicability domains of these models were determined via leverage approach. The simple developed four-parameter linear model can predict the gas chromatographic RIs of terpenols.