CascadeMAP: Autonomous Closed-loop Optimization of Enzyme Cascades via Microfluidics, Machine Learning and Agentic AI

CascadeMAP autonomously optimized enzyme cascades using microfluidics, machine learning, and agentic AI, processing approximately 295,000 reactions across 7,400 conditions in 7 days.

• Why it matters: Optimizing enzyme cascades involves exploring high-dimensional spaces of enzyme ratios and reaction conditions, which is resource-intensive and time-consuming, limiting rapid development of biocatalytic systems.
• What they did: The platform combines high-throughput microfluidics with Bayesian optimization and multi-agent AI to automate data processing, pattern recognition, and insight generation across two enzyme cascades monitored by fluorescence and Raman spectroscopy, respectively.
• The result: CascadeMAP achieved comparable optimization results with threefold fewer experiments than traditional methods, demonstrating a resource-efficient, generalizable approach that accelerates enzyme cascade and metabolic pathway development.