Research Article
Shaomin Yan and Guang Wu
Abstract
The Michaelis-Menten constant, Km, is important to understand the characteristics of enzyme and its relationship with substrates and numerous conditions in biochemical reactions. Although the fast development is evidenced in enzymatic research, the Km value in each enzyme under various conditions still needs to be measured individually. On the other hand, the modern computational techniques and bioinformatics provide the opportunity to theoretically predict Km in enzyme with different substrates under various conditions. Cellulose 1,4-beta-cellobiosidase is an enzyme used in cellulose hydrolysis for bio-fuel industry, and huge efforts are made to enhance its efficiency through searching for new strains of beta-cellobiosidase as well as enzymatic engineering. Therefore it is considered important to develop methods to predict the Km value in beta-cellobiosidase’s reaction. In this study, the information of amino acid properties in beta-cellobiosidase, pH and temperature in reaction, and lactoside as substrate were chosen as predictors to predict the Km values by feedforward backpropagation neural networks, and the delete-1 jackknife was used to validate the predictive model. The results show that 11 of 25 scanned amino acid properties could act as predictors, and that the amino-acid distribution probability appeared the best predictor. The two-layer structure of neural network configuration was sufficient for initial scanning. In consistent with previous studies, the Km value of enzymatic reactions was predictable using enzyme sequence information and reaction conditions with neural network models.