Stoichiometric identification with maximum likelihood principal component analysis

Area of Science:

• Biotechnology
• Bioprocess Engineering
• Microbial Physiology

Background:

• Accurate modeling of microbial cultures in bioreactors is crucial for process optimization.
• Existing models often require detailed kinetic information, limiting their applicability.
• Black-box approaches offer an alternative but need effective stoichiometric identification.

Purpose of the Study:

• To present an effective procedure for determining a biologically inspired, black-box model of microbial cultures.
• To identify the number of macroscopic reactions and their stoichiometry independently of kinetics.
• To evaluate the procedure using yeast growth data and compare it with existing methods.

Main Methods:

• Utilizes experimental data on extracellular species concentration changes over time.
• Employs maximum likelihood principal component analysis (MLPCA) for stoichiometric identification.
• Applies the procedure to the stoichiometric identification of Kluyveromyces marxianus growth on cheese whey.

Main Results:

• Successfully determines an appropriate number of macroscopic reactions and their stoichiometry.
• Demonstrates the procedure's effectiveness in a real-world bioprocess scenario.
• Provides a geometric interpretation of the stoichiometric matrix and discusses equivalent reaction schemes.

Conclusions:

• The developed procedure offers an effective way to model microbial cultures in bioreactors.
• MLPCA is a powerful tool for stoichiometric identification in bioprocesses.
• The method is robust and applicable across various microbial systems and substrates.