Accelerating Protein Evolution with AI and Automated Biofoundries

A new study showcases a revolutionary integration of protein language models with automated biofoundries to enhance protein evolution. Traditional methods like directed evolution are slow and labor-intensive, but this research introduces a closed-loop platform leveraging the protein language model ESM-2 for zero-shot variant predictions. The system automates the Design-Build-Test-Learn cycle, iterating through multiple evolution rounds within days. Using tRNA synthetase as a test case, the approach improved enzyme activity by 2.4-fold in just 10 days. This AI-driven biofoundry system marks a major leap in protein engineering, paving the way for rapid and precise biotechnological advancements.