AI-Driven Self-Driving Lab Accelerates Enzyme Improvement with Automated Biology

Self-Driving Lab Integrates AI and Automation to Advance Enzyme Engineering

Researchers at the University of Illinois Urbana-Champaign have developed a streamlined system that combines artificial intelligence (AI) with automated synthetic biology to enhance enzyme performance. This approach accelerates the design, production, and testing of enzyme variants, enabling improvements that were previously difficult to achieve.

Addressing Enzyme Limitations with AI

Enzymes are catalysts essential to many biological processes and have applications across energy, medicine, and consumer products like detergents. However, their widespread use is often limited by challenges in improving efficiency or specificity in complex chemical environments.

The difficulty lies in identifying beneficial mutations, which often involve multiple synergistic changes rather than single alterations. The Illinois team’s AI models predict which sequence changes could improve enzyme function, significantly narrowing down the vast number of possible variants.

Nilmani Singh, co-first author of the study, illustrates the scale of the challenge: “For a typical enzyme, the number of possible variations exceeds the number of atoms in the universe.” AI helps focus on a manageable library of promising candidates instead of random trial and error.

Integration with Automated Synthetic Biology

To generate the large datasets needed to train and refine AI models, the researchers partnered the AI design tool with the iBioFoundry’s automated protein synthesis and testing platform. This facility, directed by Professor Huimin Zhao, is equipped for rapid, user-friendly engineering of biological systems, ranging from enzymes to cells.

The process follows a loop: AI suggests enzyme sequence changes based on existing datasets, the iBioFoundry produces and tests these variants, and the experimental results feed back into the AI to improve subsequent designs. This creates a self-driving lab capable of iterative optimization with minimal manual intervention.

Demonstrated Improvements in Industrial Enzymes

Applying this method, the team enhanced two industrially important enzymes:

• An enzyme added to animal feed showed a 26-fold increase in activity, improving nutritional value.
• A catalyst used in chemical synthesis achieved 16 times greater activity and 90 times higher substrate specificity, reducing off-target reactions.

These results highlight the method’s broad applicability. “We only need a protein sequence and an assay,” Zhao explains, “making this a generalized platform for enzyme improvement.”

Future Developments and Accessibility

The researchers plan to refine their AI models and upgrade automation equipment for faster and higher-throughput synthesis and testing. They have also developed a user-friendly interface that allows users to run the system simply by typing queries in natural language. This lowers the barrier for experimental scientists unfamiliar with programming.

Graduate student Tianhao Yu notes, “Users don’t need to run Python code; they just describe their needs in English, and the system executes automatically.” This approach aims to accelerate enzyme engineering across diverse research fields, including drug development and sustainable technologies.

The work received support from the National Science Foundation and the U.S. Department of Energy. The full research paper, “A generalized platform for artificial intelligence-powered autonomous enzyme engineering”, is available online.

For additional perspectives on AI-driven automation in biological research, explore relevant courses and resources at Complete AI Training.