First authorMolecule · PROTEINS (Wiley) · 2026 · under review

ARC: Assessment of Interface Residue Conformation using an Ensemble of Graph Neural Networks

An ensemble of graph neural networks that scores how accurate a predicted protein interface really is.

Message passing over a residue contact graph — the core of the ARC ensemble.

Protein interface quality assessment (CASP16)

Abstract

ARC estimates the per-residue local accuracy of predicted protein structures using an ensemble of graph neural network architectures, developed as part of CASP16, the leading community-wide blind assessment in computational structural biology. It focuses on the quality of interface residue conformations, extending model-quality assessment to the regions where proteins meet and interact.

Authors

Shrestha B, Siciliano AJ, Huang G, Bao Y, Wang Z

PROTEINS: Structure, Function, and Bioinformatics · Manuscript ID 5704351 · under review · submitted Mar 2026

How to cite

@misc{shrestha2026arc,
  title  = {ARC: Assessment of Interface Residue Conformation using an Ensemble of Graph Neural Networks},
  author = {Shrestha, Bishal and Siciliano, A. J. and Huang, G. and Bao, Y. and Wang, Zheng},
  year   = {2026},
  note   = {Manuscript under review, PROTEINS: Structure, Function, and Bioinformatics}
}

More at the molecule scale

Inferring the qualities of protein-RNA models with graph transformers
Bioinformatics · 2026
CITED2 is a druggable epigenetic switch coupling neuronal maturation to regenerative decline
EMBO Mol. Med. · 2026

Discuss ARC