QSPR analyses for aminograms in food: Citrus juices and concentrates

Dragon theoretical descriptors were derived for a set of optimised amino acid structures, with the purpose of establishing quantitative structure–property relationship (QSPR) models to predict aminograms for 100% natural fresh juices and concentrates of Navel and Valencia oranges, and Eureka lemon. We used the statistical replacement method technique for designing the best multi-parametric linear regression models, which included structural features selected from a pool containing 1497 constitutional, topological, geometrical, or electronic types of molecular descriptors. The prediction results achieved in this work were in most cases in good agreement with experimental amino acid profiles obtained in our laboratories by a validated HPLC procedure, thus demonstrating the predictive power of the designed QSPR. The developed approach is of practical value, especially when it is not possible to assign the analyte concentration with an accurate degree of certainty.