Enzymatic Hydrolysis of Alaska Pollock Proteins Based on Kinetics Model and Lysine Biosensor–Neural Network Model

Abstract

ABSTRACTControlled proteolysis is important for target hydrolysates. The hydrolysis kinetics of pollock bone protein (PBP) was obtained by mathematic deduction and experimental analysis. The equation of degree of hydrolysis was as follows: DH = 2.801 ln|1 + (0.06044E0/S0−0.1005)t|, which could predict the hydrolysis process of PBP when hydrolysis condition was strictly controlled. However, the hydrolysis reaction was affected by several factors, and the predicted value may be against the experimental result. Therefore, online monitoring was necessary for obtaining the target peptides. Biosensor and artificial neural network (ANN) were employed for monitoring hydrolysis process online. Free lysine level was chosen as the monitoring factor determined by immobilized lysine oxidase electrodes. Based on the lysine biosensor and ANN, the hydrolysis monitoring model LYS-ANN was built, including 3 input layer nodes (E0, S0, and LYS), 1 output layer node (DH), and 11 hidden layer nodes. R2 value between sample value and simulation value was 0.9964. Simulation error was in the range of 0–4.56%, and the average relative error was 0.94%. The verification tests of PBP hydrolysis showed that the LYS-ANN model could forecast hydrolysis process successfully with high efficiency and accuracy even if the hydrolysis conditions varied.