EUVIMEDEUVIMED entdecken
Open Access · CC BY 4.0

HKD-CPI: high-order knowledge distillation enhanced inductive compound-protein interaction prediction

Zhongyu He, Xiangrong Liu iD, Yinghui Jiang, Junlin Xu, Yuan Lin, Shuting Jin iD, Leyi Wei iD, Youyu Wang

Bioinformatics · 01.06.2026 · 42 (6)

DOI: 10.1093/bioinformatics/btag290 ↗

Kurz gefasst

Fachartikel in der Zeitschrift Bioinformatics, online erschienen am 21.05.2026. Verfasst von Zhongyu He und 7 weiteren Autor:innen. Beteiligte Einrichtung: School of Informatics, Xiamen University, Xiamen, 361005, China. Hauptthemen laut MEDLINE: Proteins, Computational Biology, Protein Interaction Mapping, Drug Discovery. Die Arbeit nennt 24 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0). Gefördert von: National Science and Technology Major Project of China, National Natural Science Foundation of China, Internal Research Grants of Macao Polytechnic University.

AbstractSchlagwörterBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

Motivation Accurately identifying compound–protein interactions (CPIs) is critical for accelerating drug discovery. Recent deep learning methods have achieved impressive results, yet they primarily focus on local structures and neighborhood information, often overlooking high-order interaction patterns shared among similar molecules. Results In this paper, we propose HKD-CPI, a high-order knowledge-enhanced inductive framework designed to improve generalization to unseen compound–protein pairs. Specifically, HKD-CPI introduces a molecular graph tokenization mechanism that aligns compound molecular graph features with token embeddings from sequence-pretrained large language models (LLMs), effectively infusing sequence-derived semantics into structural representations. To capture shared interaction patterns among functionally similar biomolecules, we construct a hypergraph-based representation to model high-order relationships between feature-similar compound/protein groups and their binding partners. Furthermore, a knowledge distillation strategy is further adopted to transfer high-order interaction knowledge from the hypergraph to a lightweight student model, enabling efficient and robust CPI prediction. Extensive experiments demonstrate that HKD-CPI outperforms existing state-of-the-art methods in inductive CPI prediction tasks. In particular, it achieves an average improvement of 4.94% in AUROC and 3.64% in AUPRC over the best-performing baseline across five benchmark datasets. Availability and implementation Our code and data are available at https://github.com/Hezy618/HKD-CPI.

Schlagwörter

MeSH-Schlagwörter (MEDLINE)

ProteinsComputational BiologyProtein Interaction MappingDrug DiscoveryDeep LearningLarge Language ModelsDatabases, ProteinProtein Binding

Substanzen

Proteins

Autor:innen und Institutionen

  • School of Informatics, Xiamen University, Xiamen, 361005, China.
  • School of Computer Science and Technology, Wuhan University of Science and Technology, Wuhan, Hubei 430065, China.
  • School of Economics, Innovation, and Technology, Kristiania University College, Bergen 5022, Norway.
  • Centre for Artificial Intelligence driven Drug Discovery, Faculty of Applied Science, Macao Polytechnic University, Macao SAR 999078, China.
  • Department of Thoracic Surgery, Sichuan Academy of Medical Sciences and Sichuan Provincial People's Hospital, University of Electronic Science and Technology of China, Chengdu, Sichuan Province 610072, China.

ORCID-Profile: Xiangrong Liu, Shuting Jin, Leyi Wei

Förderung

  • National Science and Technology Major Project of China
  • National Natural Science Foundation of China
  • Internal Research Grants of Macao Polytechnic University

Zitieren

APA

He, Z., Liu, X., Jiang, Y., Xu, J., Lin, Y., Jin, S., Wei, L., & Wang, Y. (2026). HKD-CPI: high-order knowledge distillation enhanced inductive compound-protein interaction prediction. Bioinformatics, 42(6). https://doi.org/10.1093/bioinformatics/btag290

Vancouver

He Z, Liu X, Jiang Y, Xu J, Lin Y, Jin S, et al. HKD-CPI: high-order knowledge distillation enhanced inductive compound-protein interaction prediction. Bioinformatics. 2026;42(6). doi:10.1093/bioinformatics/btag290

BibTeX
@article{He2026,
  author = {He, Zhongyu and Liu, Xiangrong and Jiang, Yinghui and Xu, Junlin and Lin, Yuan and Jin, Shuting and Wei, Leyi and Wang, Youyu},
  title = {HKD-CPI: high-order knowledge distillation enhanced inductive compound-protein interaction prediction},
  journal = {Bioinformatics},
  year = {2026},
  volume = {42},
  number = {6},
  doi = {10.1093/bioinformatics/btag290}
}
Kostenlos starten →