A Customized Bayesian Algorithm to Optimize Enzyme-Catalyzed Reactions.

Area of Science:

• Chemical Engineering
• Biocatalysis
• Computational Chemistry

Background:

• Design of Experiments (DoE) is crucial for optimizing chemical reaction performance.
• Response Surface Methodology (RSM) is a common DoE approach but requires numerous experiments with increasing variables.
• Enzyme-catalyzed reactions require efficient optimization of continuous parameters.

Purpose of the Study:

• To develop and validate an improved Bayesian optimization algorithm (BOA) for optimizing enzyme-catalyzed reactions.
• To address the limitations of RSM in handling a large number of experimental variables.
• To maximize the catalytic performance of enzyme-catalyzed reactions under limited experimental resources.

Main Methods:

• Implementation of a novel Bayesian optimization algorithm (BOA).
• Optimization of continuous parameters such as temperature, reaction time, and reactant/enzyme concentrations.
• Benchmarking BOA against Response Surface Methodology (RSM) and existing Bayesian optimization algorithms.
• Validation using biocatalytic C-C bond formation and amination reactions.

Main Results:

• The proposed BOA achieved significant improvements in optimizing turnover number.
• Up to 80% improvement was observed compared to RSM.
• Up to 360% improvement was obtained compared to previous Bayesian optimization algorithms.
• Simultaneous optimization of enzyme activity and selectivity was demonstrated for cross-benzoin condensation.

Conclusions:

• The developed BOA offers a more efficient and effective approach for optimizing enzyme-catalyzed reactions.
• BOA provides superior performance over RSM and existing Bayesian methods, especially under resource constraints.
• This strategy enhances both enzyme activity and selectivity, paving the way for advanced biocatalytic applications.