Prediction of strain engineerings that amplify recombinant protein secretion through the machine learning approach...

Area of Science:

• Biotechnology and synthetic biology
• Industrial microbiology
• Computational biology

Background:

• Precision fermentation (PF) is a rapidly advancing field poised for significant growth.
• PF utilizes engineered microorganisms to produce specific molecules, with applications spanning food, pharmaceuticals, and materials.
• Understanding host organisms and strain engineering is crucial for optimizing PF processes.

Purpose of the Study:

• To provide a comprehensive overview of precision fermentation, including its history, growth projections, and market potential.
• To detail various host organisms used in PF, evaluating their respective advantages and disadvantages.
• To introduce a novel machine learning approach for predicting and optimizing strain engineering strategies in PF.

Main Methods:

• Review of precision fermentation history, applications, and market landscape.
• Description of prokaryotic and eukaryotic host organisms for PF.
• Development and validation of a machine learning model (MaLPHAS) for predicting protein secretion.
• In silico cross-validation and in vitro testing of MaLPHAS predictions.

Main Results:

• The study outlines the expected 70% growth of the PF market in the next five years.
• MaLPHAS demonstrated up to 46.6% R² accuracy in predicting strain engineering outcomes.
• In vitro testing identified a gene engineering strategy that doubled heterologous protein secretion in Komagataella phaffii, outperforming existing methods.

Conclusions:

• Precision fermentation is a key technology with substantial market disruption potential.
• Machine learning, exemplified by MaLPHAS, offers a powerful approach to accelerate PF process optimization.
• Optimized strain engineering through computational methods can significantly enhance the production of valuable recombinant proteins.