ARTFEED — Contemporary Art Intelligence

TriGlue: Generative Model for Molecular Glue-Induced Ternary Complexes

publication · 2026-07-27

A novel computational framework named TriGlue has been introduced for the creation of molecular glue degraders, facilitating the formation of ternary complexes between a target protein and an E3 ubiquitin ligase. This research, detailed in arXiv (2607.22143), explores the relatively uncharted territory of computational design for molecular glues. In contrast to traditional structure-based drug design, this method deals with unknown protein-protein interfaces and necessitates the concurrent modeling of ligand generation, docking, and ternary complex formation. TriGlue addresses this by breaking it down into two phases: estimating the interface and generating the complex conditioned on that interface. The initial phase employs an SE(3)-equivariant module for interface prediction, drawing inspiration from biological mechanisms of molecular glue activity, potentially enhancing targeted protein degradation therapies.

Key facts

  • TriGlue is a generative model for molecular glue-induced ternary complexes.
  • Molecular glue degraders induce ternary complex formation between E3 ubiquitin ligase and target protein.
  • The work is published on arXiv with ID 2607.22143.
  • Computational design of molecular glues is largely unexplored.
  • Molecular glue design requires modeling ligand generation, protein-protein docking, and ternary complex assembly.
  • TriGlue decomposes ternary complex generation into interface estimation and interface-conditioned complex generation.
  • The interface estimation module is SE(3)-equivariant.
  • The framework is motivated by the mechanism of molecular glue action.

Entities

Institutions

  • arXiv

Sources