Flagship Pioneering

Long-context co-folding model for protein, nucleic-acid and ligand assemblies, folding systems up to 16,384 residues on a single GPU.

All-atom structure tokenizer that turns proteins, RNA and small molecules into discrete 3D tokens and decodes them back below 1 Å RMSE.

Protein inverse folding by categorical diffusion, tuned with reinforcement learning on structural consistency. Samples diverse refoldable sequences.