A combination of automatic untargeted metabolic profiling analysis with targeted metabolomics to enhance geographical discrimination of Goji berry

The quality variation of Goji berry across cultivated zones leaded to an urgent requirement of geographical discrimination. Classical untargeted to targeted strategy faced with the problem of reducing prediction capability. Here, a new untargeted to targeted transition strategy was proposed. At first, the AntDAS-GCMS was used for automatically resolving compounds in untargeted metabolic profiling data to initialize a biomarker set. Compounds with similar chemical structures and/or homologous metabolic pathways were introduced to enhance the coverage of biomarker set, which were used to construct an MRM-based targeted metabolomics method. Quantitative analysis revealed that, proline was dramatically higher in Gansu (252.31 ± 310.68 μg/g) than Ningxia (1.85 ± 4.56 μg/g) and Xinjiang (0.08 ± 0.18 μg/g); cellobiose showed 4.2-fold higher concentration in Xinjiang (84.32 ± 28.54 μg/g) than Ningxia (20.32 ± 3.90 μg/g); glucose was increased in Ningxia (5255.56 ± 425.48 μg/g). The developed strategy achieved >96% prediction accuracy for Goji berries from different zones, outperforming classical untargeted-to-targeted approaches (>91%). • An untargeted to targeted strategy was developed for discriminating Goji berry. • AntDAS-GCMS enables automatic marker screening and identification in untargeted way. • Valuable compounds beyond the screened ones can be obtained with the new strategy. • GC–MS/MS-based targeted method can greatly enhance geographical discrimination. • The developed strategy outperformed the traditional untargeted to targeted method.