CAGenMol: Condition-Aware Diffusion Language Model for Goal-Directed Molecular Generation
This work addresses the challenge of reconciling conflicting objectives in molecular generation for drug discovery, offering a unified framework that improves multi-objective optimization.
CAGenMol proposes a condition-aware discrete diffusion framework for goal-directed molecular generation that integrates heterogeneous constraints via reinforcement learning, achieving consistent improvements over state-of-the-art methods in binding affinity, drug-likeness, and success rate across multiple benchmarks.
Goal-directed molecular generation requires satisfying heterogeneous constraints such as protein--ligand compatibility and multi-objective drug-like properties, yet existing methods often optimize these constraints in isolation, failing to reconcile conflicting objectives (e.g., affinity vs. safety), and struggle to navigate the non-differentiable chemical space without compromising structural validity. To address these challenges, we propose CAGenMol, a condition-aware discrete diffusion framework over molecular sequences that formulates molecular design as conditional denoising guided by heterogeneous structural and property signals. By coupling discrete diffusion with reinforcement learning, the model aligns the generation trajectory with non-differentiable objectives while preserving chemical validity and diversity. The non-autoregressive nature of diffusion language model further enables iterative refinement of molecular fragments at inference time. Experiments on structure-conditioned, property-conditioned, and dual-conditioned benchmarks demonstrate consistent improvements over state-of-the-art methods in binding affinity, drug-likeness, and success rate, highlighting the effectiveness of our framework.