In vino veritas: A metabolomics approach for authenticating Provence Rosé wines

Provence rosé wines have gained global popularity, making them vulnerable to fraud. This study aimed to identify specific chemical markers to detect counterfeit Provence rosé wines. An untargeted LC-MS-based metabolomics analysis was performed on a set of 30 wines classified as "Provence," "Non-Provence," and "Provence imitations." Using the Molnotator workflow, 1300 potential metabolites were generated, and five key chemomarkers were selected through a machine learning pipeline. Further targeted analysis and bioinformatics using in silico MS/MS fragmentation systems confidently annotated three specific chemomarkers for "Provence" rosé: acuminoside, tetrahydroxydimethoxyflavone, and 5'-methoxycastavinol. A composite score using a PLS model combining the 3 chemomarkers effectively distinguished authentic wines, with high accuracy (sensitivity 83.3 %, specificity 100 %, accuracy 93.3 %).