disease-aware

DrugGen-2 fine-tunes GPT-2 using supervised learning and reinforcement learning (GRPO) to generate small molecules conditioned on both disease ontology and target protein sequences, achieving superior diversity and binding affinity for drug discovery.