Novel two variable-QSPR analysis for authentication and typification of vegetable oils

Abstract

In the ongoing research studies on the prediction of complex chemical mixtures, the present work typifies various vegetable oils containing fatty acids in given proportions and extracted from different plant sources. This is important in the study of the properties conferred on foods, both from an organoleptic and nutritional point of view. Therefore, the ratio between the values of the saponification and iodine indices is used for the first time to carry out a Quantitative Structure-Property Relationship (QSPR) study, both parameters being the most important for characterizing oils and fats from different sources. QSPRs were formulated in 144 vegetable oils, composed of 1–8 fatty acid components. A set of 25,118 mixture descriptors was calculated as linear combinations of the non-conformational descriptors of the fatty acid components and their weight percent compositions. This approach is useful for discerning natural oils, and the Replacement Method variable subset selection technique is applied afterwards to select the best mixture descriptors in the predictive model. Finally, different vegetable oils with known composition, but unknown experimental saponification and iodine indices data, were predicted, and were successfully classified using the established QSPR. This two-variable QSPR analysis can beextended to fats and other types of oils, such as fish oils. It also serves as a background and database for other methodologies.