FRAGMENTA: End-to-end Fragmentation-based Generative Model with Agentic Tuning for Drug Lead Optimization
This work addresses drug lead optimization for medicinal chemists by automating model tuning and improving molecule diversity, though it builds incrementally on fragment-based approaches.
The paper tackled the problem of molecule generation for drug discovery with limited class-specific data by introducing FRAGMENTA, an end-to-end framework that combines a generative model with dynamic Q-learning and an agentic AI system for tuning via expert feedback, resulting in nearly twice as many high-scoring molecules as baselines in cancer drug experiments.
Molecule generation using generative AI is vital for drug discovery, yet class-specific datasets often contain fewer than 100 training examples. While fragment-based models handle limited data better than atom-based approaches, existing heuristic fragmentation limits diversity and misses key fragments. Additionally, model tuning typically requires slow, indirect collaboration between medicinal chemists and AI engineers. We introduce FRAGMENTA, an end-to-end framework for drug lead optimization comprising: 1) a novel generative model that reframes fragmentation as a "vocabulary selection" problem, using dynamic Q-learning to jointly optimize fragmentation and generation; and 2) an agentic AI system that refines objectives via conversational feedback from domain experts. This system removes the AI engineer from the loop and progressively learns domain knowledge to eventually automate tuning. In real-world cancer drug discovery experiments, FRAGMENTA's Human-Agent configuration identified nearly twice as many high-scoring molecules as baselines. Furthermore, the fully autonomous Agent-Agent system outperformed traditional Human-Human tuning, demonstrating the efficacy of agentic tuning in capturing expert intent.