Structure-activity relationships between sterols and their thermal stability in oil matrix

Structure-activity relationships between 20 sterols and their thermal stabilities were studied in a model oil system. All sterol degradations were found to be consistent with a first-order kinetic model with determination of coefficient (R2) higher than 0.9444. The number of double bonds in the sterol structure was negatively correlated with the thermal stability of sterol, whereas the length of the branch chain was positively correlated with the thermal stability of sterol. A quantitative structure-activity relationship (QSAR) model to predict thermal stability of sterol was developed by using partial least squares regression (PLSR) combined with genetic algorithm (GA). A regression model was built with R2 of 0.806. Almost all sterol degradation constants can be predicted accurately with R2 of cross-validation equals to 0.680. Four important variables were selected in optimal QSAR model and the selected variables were observed to be related with information indices, RDF descriptors, and 3D-MoRSE descriptors.