This feasibility study reports the use of direct analysis in real-time high-resolution mass spectrometry (DART-HRMS) in profiling the powders from edible insects, as well as the potential for the identification of different insect species by classification modeling. The basis of this study is the revolution that has occurred in the field of analytical chemistry, with the improved capability of ambient mass spectrometry to authenticate food matrices. In this study, we applied DART-HRMS, coupled with mid-level data fusion and a learning method, to discriminate between Acheta domesticus (house cricket), Tenebrio molitor (yellow mealworm), Locusta migratoria (migratory locust), and Bombyx mori (silk moth). A distinct metabolic fingerprint was observed for each edible insect species, while the Bombyx mori fingerprint was characterized by highly abundant linolenic acid and quinic acid; palmitic and oleic acids are the statistically predominant fatty acids in black soldier fly (Hermetia illucens). Our chemometrics also revealed that the amino acid proline is a discriminant molecule in Tenebrio molitor, whereas palmitic and linoleic acids are the most informative molecular features of the house cricket (Acheta domesticus). Good separation between the four different insect species was achieved, and cross-validation gave 100% correct identification for all training samples. The performance of the random forest classifier was examined on a test set and produced excellent results, in terms of overall accuracy, sensitivity, and specificity. These results demonstrate the reliability of the DART-HRMS as a screening method in a future quality control scenario to detect complete substitution of insect powders.
Keywords: DART-HRMS; data fusion; fingerprinting; frauds; house cricket; migratory locust; silk moth; yellow mealworm.