Summary
Designing new drug molecules that fit well into specific protein pockets is a complex challenge, requiring careful control of shape and chemical properties. The researchers propose PocketVE, a new method that uses a type of artificial intelligence called diffusion models to generate 3D drug molecules conditioned on the protein pocket. Their method improves the stability and physical realism of the generated molecules while allowing guided adjustments to desired properties without needing extra classifiers. Tests show PocketVE creates better-fitting and less strained molecules compared to previous approaches, highlighting the importance of balancing geometric stability with property guidance.
protein pocket3D molecule generationdiffusion modelsvariance-exploding diffusionstructure-based drug designcoordinate denoisingclassifier-free guidancemolecular propertiesstrain energygeometric stability
Abstract
Protein-conditioned 3D molecule generation is a central challenge in structure-based drug design, requiring a balance between pocket compatibility, molecular properties, and physical geometry. We propose \textbf{PocketVE}, a protein-pocket-conditioned variance-exploding (VE) diffusion framework that couples stable coordinate denoising with inference-time property guidance. Specifically, PocketVE combines an EDM-style training and sampling setup for 3D denoising, classifier-free guidance for multi-property steering without external property classifiers, and adaptive protein perturbation as a training-time pocket regularizer. Evaluated on CrossDocked2020 under the GenBench3D protocol, PocketVE improves Valid$_{3\text{D}}$ from 58.6 to 80.6 and reduces strain energy from 457.4 to 127.9 relative to its TAGMol architectural baseline, while retaining competitive docking and molecular-property scores under moderate guidance. A guidance-scale study shows that moderate guidance gives a favorable balance between target-related objectives and geometric quality, whereas stronger guidance can degrade geometry and distributional fidelity. Pocket-permutation and PoseCheck diagnostics further support pocket-specific spatial compatibility with reduced steric conflicts. Overall, the results suggest that geometric stability and inference-time property guidance should be considered as coupled design objectives.