🛰️ Daily AI Frontier
‹ back to 2026-07-27

TriGlue: a Biology-Inspired Generative Model for Generating Molecular Glue-Induced Ternary Complex

arXiv cs.LG Bioinformatics AI Yuliang Yan, Shuo Yan, Haochun Tang, Yiqin Sun, Enyan Dai 2026-07-24

TL;DR - TriGlue is a generative framework for designing molecular glues and their induced E3 ligase–target protein ternary complexes. It could accelerate targeted protein-degradation drug discovery by jointly addressing ligand generation, protein docking, and complex assembly.

  • Decomposes generation into protein-interface estimation and interface-conditioned ternary-complex generation.
  • Uses an SE(3)-equivariant module to predict constrained interfaces from unbound protein structures.
  • Applies flow matching to jointly generate a molecular glue and the rigid-body transformation needed for assembly.
  • Experiments report chemically valid molecules and plausible ternary complexes.

view merged work →