Direct Identification and Quantitation of Protein Peptide Powders Based on Multi-Molecular Infrared Spectroscopy and Multivariate Data Fusion.

Given that protein peptide powders (PPPs) from different biological sources were inherited with diverse healthcare functions, which aroused adulteration of PPPs. A high-throughput and rapid methodology, united multi-molecular infrared (MM-IR) spectroscopy with data fusion, could determine the types and component content of PPPs from seven sources as examples. The chemical fingerprints of PPPs were thoroughly interpreted by tri-step infrared (IR) spectroscopy, and the defined spectral fingerprint region of protein peptide, total sugar, and fat was 3600-950 cm, which constituted MIR finger-print region. Moreover, the mid-level data fusion model was of great applicability in qualitative analysis, in which the F1-score reached 1 and the total accuracy was 100%, and a robust quantitative model was established with excellent predictive capacity (: 0.9935, RMSEP: 1.288, and RPD: 7.97). MM-IR coordinated data fusion strategies to achieve high-throughput, multi-dimensional analysis of PPPs with better accuracy and robustness which meant a significant potential for the comprehensive analysis of other powders in food as well.