TriGlue: A Biology-Inspired Generative Model for Molecular Glue-Induced Ternary Complexes
Original title:TriGlue: a Biology-Inspired Generative Model for Generating Molecular Glue-Induced Ternary Complex
TriGlue formulates molecular glue design as a ternary-complex generation problem. Its two-stage framework first estimates a geometrically constrained protein-protein interface from unbound monomer structures using an SE(3)-equivariant module. An interface-conditioned ternary flow-matching network then jointly generates a molecular glue and predicts the rigid-body transformation needed to assemble the complex. The abstract reports that extensive experiments produced chemically valid molecules and plausible ternary complexes, but it does not provide quantitative benchmark results. Code is available on GitHub.
Why it's worth reading
Molecular glue discovery couples unknown protein interfaces, ligand generation, and complex assembly. TriGlue offers a unified generative formulation, making its implementation and full benchmark details especially relevant to evaluating whether this approach is practically useful.