Unraveling the potential of diffusion models in small-molecule generation

Abstract

Generative artificial intelligence (AI) presents chemists with novel ideas for drug design and facilitates the exploration of vast chemical spaces. As an emerging tool, diffusion models (DMs) have recently attracted great attention in drug research and development (R&D). Here, we comprehensively review the latest advances in, and applications of, DMs in molecular generation. We introduce the theoretical principles of DMs and then categorize various DM-based molecular generation methods according to their mathematical and chemical applications. We also examine the performance of these models on benchmark datasets, with a particular focus on comparing the generation performance of existing 3D methods. Finally, we conclude by emphasizing current challenges and suggesting future research directions to fully exploit the potential of DMs in drug discovery.