Multi-agent system accelerates drug discovery from start to finish
MADDIE: Multi-Agent Drug Discovery Engine
Artificial Intelligence
Summary
Finding new medicines usually takes a lot of time, money, and expert effort. The authors present MADDIE, a system with multiple AI agents, each responsible for a key step in drug discovery. These agents work together flexibly to design, optimize, synthesize, and test molecules. MADDIE performed well on 50 different protein targets, creating molecules that meet important criteria and outlining how to make them. It is more efficient and effective than previous single-agent or fixed-workflow approaches.
What this means in practice
- •For pharmaceutical development teams: Generate promising drug candidates and synthesis plans automatically to reduce time and expert effort.
- •For chemical manufacturing planners: Plan feasible chemical synthesis routes for novel molecules discovered through multi-agent AI workflows.
- •For biotech startup developers: Build automated platforms that accelerate drug discovery pipelines by integrating flexible multi-agent AI components.$Commercial implications: Enables creation of commercial AI-driven drug design services currently limited by inflexible or single-agent systems.
Authors
Ye Liu, Reza Averly, Avery Meyer, Frazier N. Baker, Ruoxi Gao, Daniel Adu-Ampratwum, Xia Ning
Abstract
Drug discovery remains a resource-intensive and time-consuming process, involving billions of dollars and a decade of development to advance a drug to market. While many in silico tools have accelerated this process, operating them still demands substantial time, effort, and expertise from medicinal chemists. Autonomous systems built on Large Language Models (LLMs) have emerged as a promising solution, yet current designs limit their performance. Existing single-agent systems scale poorly to long-horizon, multi-step, and realistic tasks, whereas multi-agent systems typically execute predefined workflows that cannot capture the iterative, non-linear nature of a real drug discovery campaign. Here, we present MADDIE, a multi-agent drug discovery engine. It comprises a central agent, Orchestrator, and four specialized worker agents: HitIdentifier, LeadOptimizer, Synthesizer, and Evaluator, each representing a critical step in a drug discovery campaign. Under Orchestrator's orchestration and coordination, these worker agents are composed into flexible workflows, enabling MADDIE to adapt its strategy and navigate the drug discovery pipeline end-to-end. Across 50 protein targets, MADDIE produces molecules satisfying key property requirements and their synthesis routes, outperforming existing agentic systems, generative models, and general-purpose LLMs by a wide margin. Our analysis reveals how these agents collaborate and that MADDIE can employ different strategies across targets to achieve the campaign's goal, while still being efficient in time and tokens. Together, they demonstrate MADDIE as a low-cost, effective, and efficient autonomous engine for accelerating drug discovery. Code and dataset is available via https://github.com/ninglab/MADDIE.