INIFTA   05425
INSTITUTO DE INVESTIGACIONES FISICO-QUIMICAS TEORICAS Y APLICADAS
Unidad Ejecutora - UE
artículos
Título:
Quantitative structure-activity relationships to predict sweet and non‑sweet tastes
Autor/es:
CRISTIAN ROJAS; PIERCOSIMO TRIPALDI; VIVIANA CONSONNI; ROBERTO TODESCHINI; DAVIDE BALLABIO; ANDREA MAURI
Revista:
THEORETICAL CHEMISTRY ACCOUNTS
Editorial:
SPRINGER
Referencias:
Lugar: Berlin; Año: 2016 vol. 135 p. 1 - 13
ISSN:
1432-881X
Resumen:
The aim of this work was the calibration andvalidation of mathematical models based on a quantitativestructure?activity relationship approach to discriminatesweet, tasteless and bitter molecules. The sweet-tasteless andthe sweet-bitter datasets included 566 and 508 compounds,respectively. A total of 3763 conformation-independentDragon molecular descriptors were calculated and subsequentlyreduced through both unsupervised reduction andsupervised selection coupled with the k-nearest neighborsclassification technique. A model based on nine descriptorswas retained as the optimal one for sweet and tastelessmolecules, while a model based on four descriptors wascalibrated for the sweetness-bitterness dataset. Models wereproperly validated through cross-validation and external testsets. The applicability domain of models was investigated,and the interpretation of the role of the molecular descriptorsin classifying sweet and non-sweet tastes was evaluated. Theclassification and the performance of the models presentedin this paper are simple but accurate. They are based on arelatively small number of descriptors and a straightforwardclassification approach. The results presented here indicatethat the proposed models can be used to accurately selectnew compounds as potential sweetener candidates.