Using generative AI to transform peptide hits into small molecule leads

• Joshua Mills and
• Yu Heng Lau

Beilstein J. Org. Chem. 2026, 22, 672–679, doi:10.3762/bjoc.22.51

• , de novo peptide binder generation, diffusion models for generating novel small molecule scaffolds, and deep-learning predictors of binding affinity to rapidly triage candidates. Keywords: diffusion models; drug discovery; generative AI; peptides; small molecules; Introduction In drug discovery

• David Baker and co-workers at the Institute of Protein Design in 2022–23 popularised the use of diffusion models to build de novo protein backbones followed by sequence design. As with structure prediction, there has since been an explosion of new methods improving upon these initial methods (Table 1

• reach parity with traditional peptide discovery methods in the future. Diffusion models to generate small molecule mimics The reliable design of synthesisable small molecules that can accurately mimic a peptide pharmacophore remains the core challenge in the overall peptide to small molecule workflow