A triphenylmethane reductase derived from Citrobacter sp. KCTC 18061P was coupled with a glucose 1-dehydrogenase from Bacillus sp. ZJ to construct a cofactor self-sufficient bienzyme biocatalytic system for dye decolorization. Fed-batch experiments showed that the system is robust to maintain its activity after 15 cycles without the addition of any expensive exogenous NADH. Subsequently, three different machine learning approaches, including multiple linear regression (MLR), random forest (RF), and artificial neural network (ANN), were employed to explore the response of decolorization efficiency to the variables of the bienzyme system. Statistical parameters of these models suggested that a three-layered ANN model with six hidden neurons was capable of predicting the dye decolorization efficiency with the best accuracy, compared with the models constructed by MLR and RF. Weights analysis of the ANN model showed that the ratio between two enzymes appeared to be the most influential factor, with a relative importance of 54.99% during the decolorization process. The modeling results confirmed that the neural networks could effectively reproduce experimental data and predict the behavior of the decolorization process, especially for complex systems containing multienzymes.
Keywords: artificial neural network; cofactor regeneration; dye decolorization; glucose 1-dehydrogenase; modeling; multiple linear regression; random forest; triphenylmethane reductase.