Mathematical modeling and optimization of cellulase protein production using Trichoderma reesei RL-P37

Area of Science:

• Biotechnology
• Biochemical Engineering
• Enzyme Kinetics

Background:

• Cellulase enzyme production by Trichoderma reesei is crucial for biomass-to-ethanol processes.
• Understanding the dynamics of fungal growth and enzyme production is key for efficient large-scale fermentation.
• Existing kinetic models may not fully capture the complex system dynamics.

Purpose of the Study:

• To develop and compare mathematical models for predicting Trichoderma reesei growth and cellulase production.
• To evaluate the efficacy of neural network parameter function modeling against traditional kinetic models.
• To optimize process conditions using the developed models.

Main Methods:

• Utilized neural network parameter function modeling, integrating neural networks with process knowledge.
• Developed complementary kinetic models for comparison.
• Estimated model parameters using laboratory data from bench-scale fermentations on lactose and xylose.

Main Results:

• The neural network-based model achieved approximately 33% lower root-mean-squared error (RMSE) in protein predictions compared to kinetic models.
• Total RMSE was reduced by about 40% with the neural network model.
• Optimizing performance indices using the neural network model resulted in 67% and 40% lower RMSE compared to kinetic models.

Conclusions:

• Neural network parameter function modeling offers a robust "macromodeling" technique for rapid dynamic model development.
• The neural network-based model provides more accurate predictions and optimization results closer to experimental data.
• This advanced modeling approach enhances the understanding and efficiency of cellulase production for biotechnological applications.