Rational protein engineering using an omni-directional multipoint mutagenesis generation pipeline

Area of Science:

• Protein Engineering
• Computational Biology
• Biotechnology

Background:

• Generative models accelerate protein design by creating large datasets.
• Identifying functional sequences within these datasets is a key challenge.

Purpose of the Study:

• To develop a novel pipeline for protein engineering using generative models.
• To enhance protein properties like thermostability and enzymatic activity.

Main Methods:

• Developed the Omni-Directional Multipoint Mutagenesis (ODM) model based on a pre-trained protein BERT model.
• Generated 100,000 mutant protein sequences.
• Ranked sequences using mutation effect prediction and developed specific models for thermostability and lysozyme activity.

Main Results:

• 62.5% of protease mutants showed improved thermostability after two design cycles.
• 50% of lysozyme mutants displayed increased bacteriolytic activity.
• The pipeline effectively identified and engineered proteins with enhanced functions.

Conclusions:

• The integrated pipeline successfully engineers proteins with desired functional enhancements.
• This approach offers a powerful strategy for de novo protein design and optimization.