The impact of different energy-saving roasting curves on the metabolomics and sensory profile of Robusta coffee: A multi-omics data fusion approach

This study investigates the impact of roasting speed on the chemical composition, sensory attributes, and energy consumption of Coffea canephora using an integrated metabolomics approach. Four Robusta coffees, processed by wet and dry methods, were roasted at fast, medium, and slow speeds to a medium roast degree. Untargeted metabolomics (LC-HRMS and GC × GC–MS) revealed distinct signatures, including polyphenols, lipids, and Maillard reaction products such as pyridines and pyrazines. Acrylamide levels were consistently below the European benchmark of 400 μg/kg under all roasting conditions. Multi-omics integration through DIABLO modelling distinguished fast- and slow-roasted coffees, linking chemical markers to sensory descriptors. Different aroma descriptors were also modulated by the roasting speed, with caramel, roasted, and green/vegetative flavours achieving stronger correlations. Interestingly, fast roasting emerged as the most sustainable method, leading to a significant reduction in natural gas consumption. Our results pave the way towards sustainability-oriented quality approaches in coffee roasting.