Metabolic Profiling of Fish Meat by GC-MS Analysis, and Correlations with Taste Attributes Obtained Using an Electronic Tongue

To evaluate the taste of ordinary muscle from white-fleshed fish, we used GC-MS metabolomic analysis to characterise the compounds therein, and correlated the obtained data with taste measurements from an electronic tongue. Prediction models using orthogonal partial least squares were produced for different taste attributes, and the primary metabolic components correlated with the taste attributes were identified. Clear differences were observed in the component profiles for different fish species. Using an electronic tongue, differences in tastes were noted among the fish species in terms of sourness, acidic bitterness, umami and saltiness. The obtained correlations allowed the construction of good taste prediction models, especially for sourness, acidic bitterness and saltiness. Compounds such as phosphoric acid, lactic acid and creatinine were found to be highly correlated with some taste attributes. Phosphoric acid in particular showed the highest variable important for prediction (VIP) scores in many of the taste prediction models, and it is therefore a candidate marker to evaluate the tastes of white-fleshed fish.