FROGENT: An End-to-End Full-process Drug Design Agent
The Drug Design Agent consists of data layer, tools layer, and applications layer.
The rapid evolution of Artificial Intelligence (AI), particularly the progress in Large Language Models (LLMs), is initiating a transformative shift in pharmaceutical research. However, the current AI landscape for drug discovery remains highly fragmented, with powerful tools often existing as isolated web applications, standalone software, or specialized code libraries. This siloed structure imposes significant operational friction, requiring researchers to possess specialized technical skills to manually manage incompatible interfaces and connect the output of one tool to the input of another. This lack of interoperability and automation not only undermines efficiency but also obstructs a holistic, end-to-end understanding of drug candidates.
To address these challenges, we introduce FROGENT, a unified, full-process drug design agent engineered to accelerate and integrate the complete drug discovery workflow. Our proposed agentic framework utilizes a Large Language Model and the Model Context Protocol to seamlessly orchestrate diverse and dynamic biochemical databases, extensible computational tool libraries, and task-specific AI models. FROGENT is designed to autonomously execute complex, multi-stage discovery pipelines, from initial target identification and virtual screening to de novo molecular design and retrosynthetic planning. By leveraging modular tool orchestration and language-based interaction, FROGENT significantly reduces technical barriers and enhances productivity, transforming the fragmented series of manual tasks into a cohesive, automated, and intelligent workflow.
