Uso de la metabolómica para verificar autenticidad, origen y aporte nutracéutico entre cultivares de Tropaeolum tuberosum

Abstract

The present study consisted of the processing and analysis of data from a metabolomics study using the nuclear magnetic resonance technique of six different cultivars of Tropaeolum tuberosum (mashua), whose results are published by Valle-Parra et al 2020: From morphological traits to the food fingerprint of Tropaeolum tuberosum through metabolomics by NMR. Data includes: cultivar ID (A-F); sample collection area (I, II and III); information on the content of 34 metabolites (in mM); and integration values of each resonance spectrum at an interval of 0.4 ppm (234 values per spectrum). These data were subjected to multivariable analysis by PCA revealing a grouping between cultivars A-B, C-D-E and F. These three subgroups coincided with the grouping of the analysis of the collection area. Subsequently, the performance of the machine learning classification models was evaluated: LDA, SVM and RF. The former had the lowest performance with a percentage accuracy range between 83.2 and 87.6, the second between 86.5 and 96.4, and RF had the best performance between 96.4 and 100. Finally, the nutraceutical contribution of the cultivars was evaluated, and they were ranked according to their content of amino acids, carbohydrates, organic acids, among other nutrients. In global and amino acid terms, cultivar F stood out for its superior contribution to the other cultivars; cultivar A has the highest content of organic acids, while cultivars B and D of carbohydrates.