Abstract
Designing molecules for flexible protein pockets poses a substantial challenge in structure-based drug discovery, as proteins often undergo conformational changes upon ligand binding. While deep learning-based methods have shown promise in molecular generation, they typically treat protein pockets as rigid structures, limiting their ability to capture the dynamic nature of protein-ligand interactions. Here, we present YuelDesign, a diffusion-based framework that jointly models the pocket structures and ligand conformations of protein-ligand complexes. YuelDesign uses E3former to maintain rotational and translational equivariance. The framework incorporates dual diffusion processes, an elucidated diffusion model (EDM) for coordinates and a discrete denoising diffusion probabilistic model (D3PM) for ligand atom types, enabling iterative refinement of both geometry and chemical identity. Our results demonstrate that YuelDesign generates molecules with favorable drug-likeness, low synthetic complexity, diverse chemical functional groups, and docking energies comparable to native ligands. YuelDesign presents a versatile framework for designing drugs in flexible protein pockets, with promising implications for drug discovery applications.