Determination of Branched-Amino Acid Content in Fermented Cordyceps sinensis Mycelium by Using FT-NIR Spectroscopy Technique

This research investigated the feasibility of Fourier transform near-infrared (FT-NIR) spectral technique in determining branched-amino acid contents (Leucine, Isolecucine, and Valine) in fermented Cordyceps sinensis (C. sinensis) mycelium. A comparison of two variable selection methods, namely competitive adaptive reweighted sampling (CARS) and genetic algorithms, showed that CARS achieved better performance calibrating by partial least square regression or least squares support vector machine methods, with the effective wavelengths mainly concentrated around 12,010–10,050, 7,500–6,000, and 5,000–4,000 cm−1. For the quantitative determination of Val, Ile, and Leu by the established models, root mean square error of prediction were 0.0662, 0.0594, and 0.0678; correlation coefficients were 0.80, 0.83, and 0.90; residual predictive deviation were 1.66, 1.78, and 2.30, respectively. This study showed that the FT-NIR spectral technique has a potential to determine branched chain amino acids in fermented C. sinensis mycelium and further to facilitate quality control during its production.