Skip to main content
Design amino acid sequences predicted to fold into a given three-dimensional backbone. These tools solve the inverse of structure prediction: given a target structure, they propose sequences that should adopt it, and score how compatible a sequence is with a backbone. Design can be conditioned on fixed positions and on any bound ligands, metals, or nucleic acids.
  • Input: a target 3D backbone, optionally with fixed positions or bound ligands, metals, and nucleic acids to respect.
  • Output: one or more sequences predicted to fold into that backbone, each with a compatibility (likelihood) score.
Meta AI Biohub
Institute for Protein Design
Institute for Protein Design