KINETICS OF THERMOSTABLE XYLANASE PRODUCTION IN MELANOCARPUS ALBOMYCES: A DEEP LEARNING AND MACHINE LEARNING PERSPECTIVE

Abstract

Melanocarpus Albomyces a thermophilic fungus known for its ability to produce thermostable xylanase enzymes. Due to the significant variation in the organism’s genome and limitations in experimentation, this study used a combined classical, deep, and machine learning approach. The study applied classic enzyme kinetic models, including Michaelis–Menten (MM), Lineweaver–Burk (LB), Eadie–Hofstee (EH), and Hanes–Woolf (HW), to estimate the kinetic constants Vmₐₓ and Km. Deep learning model used to develop the synthetic data and observed that three-layer Artificial Neural Network (ANN) achieved results, with an R² of 0.996 and RMSE of 18.72 µmol/L·min, showing a strong fit with lab data. In addition, Machine learning methods, SVM, KNN, Logistic Regression (LR), Random Forest (RF), and XGBoost were tested with kinetic data. Random Forest reached an accuracy of 96.5%. GridSearchCV was used to refine the hyperparameters, leading to XGBoost achieving the best cross-validated accuracy of 0.981.