Context-informed subgraph foundation models enable interpretable protein-function prediction

doi:10.1016/j.cels.2026.101535

Context-informed subgraph foundation models enable interpretable protein-function prediction

2026-03-19

0

PRE

AI

DOI：10.1016/j.cels.2026.101535

Original

Original request for help

Save

Abstract

En 中文

• We construct a knowledge graph integrating protein interactions, structures, and GO terms • We develop an inductive subgraph model for context-aware protein-function prediction • Our model excels in zero-shot, few-shot, and low-homology function prediction • Our approach complements sequence- and structure-based annotation methods

Keywords:

protein interaction

knowledge graph

subgraph model

protein-function prediction

zero-shot learning

Context-informed subgraph foundation models enable interpretable protein-function prediction

Abstract

Journal

Organization