RAG-GNN: Integrating Retrieved Knowledge with Graph Neural Networks for Precision Medicine

arXiv:2602.00586 · q-bio.MN, cs.AI, cs.LG · Submitted 2026-01-31 · Read on arXiv

q-bio.MN, cs.AI, cs.LG

Submitted: 2026-01-31

Updated: 2026-05-13

Journal ref: Front. Artif. Intell., 10 July 2026, Sec. Medicine and Public Health

DOI: 10.3389/frai.2026.1851917

Code: https://github.com/HasiHays/RAGGNN

License: http://creativecommons.org/licenses/by/4.0/

The gist: Network topology excels at structural predictions but fails to capture functional semantics encoded in biomedical literature.

Terminology

Abstract

Network topology excels at structural predictions but fails to capture functional semantics encoded in biomedical literature. We present RAG-GNN, an end-to-end trainable retrieval-augmented graph neural network framework that integrates GNN representations with dynamically retrieved literature-derived knowledge through a jointly optimized retrieval projection, gated fusion mechanism, and contrastive alignment. In a cancer signaling case study (379 proteins, 3,498 interactions, 14 functional categories), RAG-GNN improves functional clustering from silhouette = -0.237 plus or minus 0.065 (GNN-only) to-0.144 plus or minus 0.066, a consistent improvement of +0.093 plus or minus 0.022 across 10 random seeds, while the learned retrieval achieves mean precision@10 = 0.242, a 152% improvement over the random baseline (0.096). Heuristic information decomposition with bootstrap confidence intervals reveals that topology and retrieval encode overwhelmingly shared information (95.6%), with retrieval improving both intra-cluster cohesion (silhouette) and cluster agreement (ARI +0.021 plus or minus 0.015). Counterfactual experiments confirm that adversarial, absent, and random retrieval all degrade performance, validating that the gated fusion mechanism depends on document content. Benchmarking against eight established embedding methods demonstrates task-specific complementarity: topology-focused methods achieve strong link prediction, while retrieval augmentation consistently improves functional clustering within the controlled GNN-only ablation. DDR1 subnetwork analysis provides confirmatory validation consistent with established synthetic lethality relationships. These results establish that topology-only and retrieval-augmented approaches serve complementary purposes for precision medicine applications.

Sources

Related papers