Biomass soft sensor for a Pichia pastoris fed-batch process based on phase detection and hybrid modeling

Area of Science:

• Biotechnology
• Bioprocess Engineering
• Microbial Fermentation

Background:

• Pichia pastoris is widely used for recombinant protein production.
• The standard bioprocess involves distinct glycerol (biomass) and methanol (protein production) phases.
• Accurate biomass monitoring is crucial for optimizing fermentation yields and quality.

Purpose of the Study:

• To establish a soft sensor for real-time biomass concentration prediction in Pichia pastoris.
• To develop a model that automatically adapts to the different phases of the bioprocess.
• To enhance process control and quality assurance in recombinant protein production.

Main Methods:

• A hybrid modeling approach combining mechanistic (carbon balance) and data-driven (multiple linear regression) techniques.
• A multilevel phase detection algorithm using off-gas CO2 and base feed data.
• Dynamic adaptation of model parameters to distinct process phases.

Main Results:

• The soft sensor achieved a mean relative prediction error of 5.52%.
• The model demonstrated a high coefficient of determination (R²) of 0.96 across the entire process.
• The soft sensor was successfully implemented for online monitoring.

Conclusions:

• The developed soft sensor provides accurate and adaptive online biomass monitoring for Pichia pastoris.
• This tool can significantly improve quality control and serve as input for advanced process control strategies.
• The hybrid modeling approach offers a robust solution for complex bioprocess monitoring.