EUVIMEDOpdag EUVIMED
Open access

ContrastQA: A label‐guided graph contrastive learning‐based approach for protein complex structure quality assessment

Lei Zhang iD, Rui Ding, Xiao Chen iD, Jie Hou iD, Dong Si iD, Yang Wang, Keying Lin, Renzhi Cao iD

Protein Science · 2026

DOI: 10.1002/pro.70734 ↗

At a glance

Journal article in Protein Science, 2026. Written by Lei Zhang and 7 co-authors. Institution: School of Computer Science and Technology, Anhui University, Hefei, China. Major MEDLINE topics: Proteins, Computational Biology. Cited 1 times so far (Crossref); 48 references. The full text is freely available in PubMed Central. Funded by: Natural Science Foundation of Anhui Province, Key Projects of the University Excellent Talents Support Plan of Anhui Provincial Department of Education.

AbstractSubjectsBibliografiske oplysningerCiteRelated studies

Abstract (original language)

Despite recent progress, the Estimation of Model Accuracy (EMA) for protein complexes remains less advanced compared to that for protein monomers. A key challenge lies in effectively integrating both interface‐specific and global structural information to accurately assess the quality of protein complexes. Here, we introduce ContrastQA, the first EMA framework for protein complexes that incorporates the proposed label‐guided graph contrastive learning based on interface quality. By integrating a geometric graph neural network to model global structural features, ContrastQA effectively captures both local (interface‐level) and global (structure‐level) information for accurate model quality estimation. ContrastQA achieved ranking losses of 0.123 and 0.116 on the TMscore and GDT‐TS metrics on the CASP16 dataset, which are 0.015 (10.9%) and 0.012 (8.7%) lower than the second‐best EMA method with ranking losses of 0.138 and 0.128. Our study demonstrates the strong effectiveness of the label‐guided graph contrastive learning module, particularly in selecting high‐quality models. These findings suggest that our graph contrastive learning framework serves as a valuable pre‐training strategy for learning protein structure representations.

Subjects

MeSH terms (MEDLINE)

ProteinsComputational BiologyGraph Neural NetworksProtein ConformationModels, MolecularDatabases, Protein

Author keywords

deep learninggraph contrastive learningprotein complexesprotein interfaceprotein quality assessmentsamples selection strategy

Substances

Proteins

Authors and institutions

  • School of Computer Science and Technology, Anhui University, Hefei, China.
  • Department of Mathematics and Computer Science, Hamilton College, Clinton, New York, USA.
  • Department of Computer Science, Saint Louis University, St. Louis, Missouri, USA.
  • School of Computing and Software, Washington Bothell University, Bothell, WA, USA.
  • School of Computer Science, Pacific Lutheran University, Tacoma, WA, USA.

ORCID profiles: Lei Zhang, Xiao Chen, Jie Hou, Dong Si, Renzhi Cao

Funding

  • Natural Science Foundation of Anhui Province
  • Key Projects of the University Excellent Talents Support Plan of Anhui Provincial Department of Education

Cite

APA

Zhang, L., Ding, R., Chen, X., Hou, J., Si, D., Wang, Y., Lin, K., & Cao, R. (2026). ContrastQA: A label‐guided graph contrastive learning‐based approach for protein complex structure quality assessment. Protein Science. https://doi.org/10.1002/pro.70734

Vancouver

Zhang L, Ding R, Chen X, Hou J, Si D, Wang Y, et al. ContrastQA: A label‐guided graph contrastive learning‐based approach for protein complex structure quality assessment. Protein Science. 2026. doi:10.1002/pro.70734

BibTeX
@article{Zhang2026,
  author = {Zhang, Lei and Ding, Rui and Chen, Xiao and Hou, Jie and Si, Dong and Wang, Yang and Lin, Keying and Cao, Renzhi},
  title = {ContrastQA: A label‐guided graph contrastive learning‐based approach for protein complex structure quality assessment},
  journal = {Protein Science},
  year = {2026},
  doi = {10.1002/pro.70734}
}
Start for free →