3D molecular generation models expand chemical space exploration in drug design

Abstract

Drug discovery is essential in human diseases but faces challenges because of the vast chemical space. Molecular generation models have become powerful tools to accelerate drug design by efficiently exploring chemical space. 3D molecular generation has gained popularity for explicitly incorporating spatial structural information to generate rational molecules. Herein, we summarize and compare common data sets, molecular representations, and generative strategies in 3D molecular generation. We also present case studies utilizing generative modeling for ligand design and outline future challenges in developing and applying 3D models. This work provides a reference for drug design researchers interested in 3D generative modeling.