Generalized AI-Powered Platform Revolutionizes Autonomous Enzyme Engineering

Laboratory experiments are the backbone of scientific research, traditionally driven by skilled researchers. However, manual tasks are time-consuming, prone to reproducibility issues, and limited in scalability. The integration of artificial intelligence (AI) and robotics in autonomous experimentation is poised to revolutionize this process.

AI-enabled systems can explore vast, multi-dimensional spaces more efficiently than traditional computational techniques, while robotics and automation can perform experiments faster, more reliably, and at higher throughputs with better scalability. This combination holds tremendous potential for diverse fields such as synthetic biology, chemical synthesis, and materials discovery.

Early demonstrations of autonomous systems include the Robot Scientist “Adam,” which autonomously generated and tested functional genomics hypotheses for Saccharomyces cerevisiae. Coscientist, a multi-LLMs-based intelligent agent, can autonomously perform chemical synthesis reactions. In materials science, the Ada platform optimized thin-film compositions, and the A-Lab combined robotics, machine learning, and historical data to synthesize inorganic powders, creating 41 new compounds in just 17 days of continuous operation.

In biology, autonomous experimentation is less mature, making even routine processes like DNA assembly, gene editing, or metabolic engineering challenging. Integrating instruments for continuous experimentation requires skilled personnel with expertise in biological experimentation, robotics, and programming. Much previous work in autonomous synthetic biology has been highly specific, targeting singular goals like engineering a single protein, metabolic engineering for a single product, or orphan enzyme identification.

To address these challenges, a new platform uses protein engineering as a case study to establish a roadmap for generalized autonomous experimentation in synthetic biology. Protein engineering provides an extensive toolkit for modifying enzymes for widespread application in fields such as medicine, biofuels, and biocatalysis. By iterative design, build, test, and learn (DBTL) cycles, enzymes can be made more stable, selective, or efficient. Methods like directed evolution and computer-aided design offer diverse strategies for protein engineering, and high-throughput screening strategies allow iterating over a large sample space.

Despite the wide applicability of protein engineering, there remain unmet needs, particularly in efficiently navigating vast sequence spaces and optimizing protein function in complex environments. The new platform aims to fill these gaps, making it more scalable and adaptable for diverse problems across different locations. This generalizable approach enables researchers to use a common scientific framework, promoting collaboration and efficient knowledge transfer while minimizing the need to redevelop similar methods for common goals.

With scalable, generalizable platforms, synthetic biology can move beyond isolated successes and drive innovation in a wide range of fields. The integration of AI and robotics in autonomous experimentation is set to transform scientific research, making it more efficient, reliable, and scalable.