Abstract
The feasibility of raw data acquired by liquid chromatography – high resolution mass spectrometry for the separation of four tequila corresponding to different maturation stages was demonstrated using partial least squares regression (PLS). Five samples of silver, rested, aged, and extra-aged tequilas from different manufacturers were analyzed each in two technical replicates. Prior to the analysis, the samples were four-fold diluted with 0.1% trifluoroacetic acid. Rectangle bucketing was performed for raw data, with a retention time width of 60 s and m/z width of 1. The prepared dataset was submitted to principal component analysis (PCA) which allowed for the detection of tequila contaminants identified to be N-lauryl diethanolamine and phthalate esters. Next, PLS was performed on depurated data, annotating tequila categories as follows: 0 - silver, 1 - rested, 2 - aged and 3 - extra aged. In the computed model, two principal components accounted for 64% and 98% of X-data and Y-data variability, respectively. Cross-validation revealed similar analytical performance in calibration and in prediction with root mean square errors of 0.1851 and 0.2082, respectively. The assignation of category’s numerical value (0-3) was attained for eight randomly selected tequilas (two per category). Statistical differences between predicted values were obtained for each pair of categories except for the aged versus extra-aged liquors (ANOVA, p < 0.05). Overall, the assignation of tequila category was achieved with no need for the detection/identification nor quantification of individual compounds.
Acknowledgements
The authors thank the Tequila Regulatory Council (Guadalajara, Mexico) for providing tequila samples from different manufacturers.
Disclosure statement
The authors report there are no competing interests to declare.