A sequence embedding method for enzyme optimal condition analysis.

Area of Science:

• Biochemistry and Molecular Biology
• Computational Biology
• Enzyme Engineering

Background:

• Enzyme activity is highly sensitive to environmental factors like pH.
• Traditional methods for optimizing enzyme conditions involve extensive experimental screening and protein engineering.
• Predicting and enhancing enzyme performance in specific environments is crucial for biotechnological applications.

Purpose of the Study:

• To develop a sequence-based computational method for predicting enzyme optimal conditions.
• To represent amino acid and structural information as informative vectors.
• To provide accurate mutation suggestions for improved enzyme activity in desired conditions.

Main Methods:

• Developed an embedding method to represent amino acids and structural information as latent space vectors.
• Utilized a probabilistic approximation method for embedding learning on glycoside hydrolase GH11 family sequences (125 samples).
• Designed a computational score based on embedding vectors to assess optimal conditions and predict mutations.

Main Results:

• Achieved 80% accuracy in predicting optimal conditions for test proteins within the GH11 family.
• The embedding vectors capture correlations between amino acids, sites, and optimal conditions.
• Mutation suggestions were consistent with prior experimental findings.

Conclusions:

• A novel computational method enables sequence-based analysis of enzyme optimal conditions.
• This efficient method reduces the need for extensive wet experiments and multiple mutations.
• Provides targeted recommendations for amino acid substitutions to achieve desired enzyme properties.