BoltzGen

Designing antibodies and other drugs in silico is a grand challenge with the potential to deliver improved controllability and efficiency over traditional lab-based screening.

BoltzGen addresses a key part of this challenge. Building on recent work in structure prediction including AlphaFold-3 and Boltz, BoltzGen is a diffusion model that generates proteins predicted to bind a given target. The target molecule can be another protein, a small molecule, or a nucleic acid. In addition to drug discovery, BoltzGen is also potentially applicable to the development of improved diagnostics and research reagents.