EUVIMEDEUVIMED entdecken
Open Access · CC BY 4.0

Topology-driven negative sampling enhances generalizability in protein–protein interaction prediction

Ayan Chatterjee iD, Babak Ravandi iD, Parham Haddadi iD, Naomi H Philip iD, Mario Abdelmessih iD, William R Mowrey iD, Piero Ricchiuto iD, Yupu Liang, Wei Ding, Juan Carlos Mobarec iD, Tina Eliassi-Rad

Bioinformatics · 2025

DOI: 10.1093/bioinformatics/btaf148 ↗

Kurz gefasst

Fachartikel in der Zeitschrift Bioinformatics, 2025. Verfasst von Ayan Chatterjee und 10 weiteren Autor:innen. Beteiligte Einrichtung: BioClarity AI, Boston, MA 02130, United States. Hauptthemen laut MEDLINE: Protein Interaction Mapping, Machine Learning, Proteins, Computational Biology. Die Arbeit wurde bisher 7-mal zitiert (Crossref) und nennt 126 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0). Gefördert von: Department of Bioinformatics and Data Sciences at Alexion AstraZeneca Rare Disease, Boston, USA.

AbstractSchlagwörterBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

Motivation Unraveling the human interactome to uncover disease-specific patterns and discover drug targets hinges on accurate protein–protein interaction (PPI) predictions. However, challenges persist in machine learning (ML) models due to a scarcity of quality hard negative samples, shortcut learning, and limited generalizability to novel proteins. Results In this study, we introduce a novel approach for strategic sampling of protein–protein noninteractions (PPNIs) by leveraging higher-order network characteristics that capture the inherent complementarity-driven mechanisms of PPIs. Next, we introduce Unsupervised Pre-training of Node Attributes tuned for PPI (UPNA-PPI), a high throughput sequence-to-function ML pipeline, integrating unsupervised pre-training in protein representation learning with Topological PPNI (TPPNI) samples, capable of efficiently screening billions of interactions. By using our TPPNI in training the UPNA-PPI model, we improve PPI prediction generalizability and interpretability, particularly in identifying potential binding sites locations on amino acid sequences, strengthening the prioritization of screening assays and facilitating the transferability of ML predictions across protein families and homodimers. UPNA-PPI establishes the foundation for a fundamental negative sampling methodology in graph machine learning by integrating insights from network topology. Availability and implementation Code and UPNA-PPI predictions are freely available at https://github.com/alxndgb/UPNA-PPI.

Schlagwörter

MeSH-Schlagwörter (MEDLINE)

Protein Interaction MappingMachine LearningProteinsComputational BiologyHumansBinding SitesDatabases, ProteinProtein Binding

Substanzen

Proteins

Autor:innen und Institutionen

  • BioClarity AI, Boston, MA 02130, United States.
  • Bioinformatics and Data Science, Alexion AstraZeneca Rare Disease, Boston, MA 02210, United States.
  • Network Science Institute, Northeastern University, Boston, MA 02115, United States.
  • Department of Physics, Northeastern University, Boston, MA 02115, United States.
  • Protein Structure and Biophysics, Discovery Sciences, R&D, AstraZeneca, Cambridge, UK.
  • Khoury College of Computer Sciences, Northeastern University, Boston, MA CB2 0AA, United States.

ORCID-Profile: Ayan Chatterjee, Babak Ravandi, Parham Haddadi, Naomi H Philip, Mario Abdelmessih, William R Mowrey, Piero Ricchiuto, Juan Carlos Mobarec

Förderung

  • Department of Bioinformatics and Data Sciences at Alexion AstraZeneca Rare Disease, Boston, USA

Zitieren

APA

Chatterjee, A., Ravandi, B., Haddadi, P., Philip, N. H., Abdelmessih, M., Mowrey, W. R., Ricchiuto, P., Liang, Y., Ding, W., Mobarec, J. C., & Eliassi-Rad, T. (2025). Topology-driven negative sampling enhances generalizability in protein–protein interaction prediction. Bioinformatics. https://doi.org/10.1093/bioinformatics/btaf148

Vancouver

Chatterjee A, Ravandi B, Haddadi P, Philip NH, Abdelmessih M, Mowrey WR, et al. Topology-driven negative sampling enhances generalizability in protein–protein interaction prediction. Bioinformatics. 2025. doi:10.1093/bioinformatics/btaf148

BibTeX
@article{Chatterjee2025,
  author = {Chatterjee, Ayan and Ravandi, Babak and Haddadi, Parham and Philip, Naomi H and Abdelmessih, Mario and Mowrey, William R and Ricchiuto, Piero and Liang, Yupu and Ding, Wei and Mobarec, Juan Carlos and Eliassi-Rad, Tina},
  title = {Topology-driven negative sampling enhances generalizability in protein–protein interaction prediction},
  journal = {Bioinformatics},
  year = {2025},
  doi = {10.1093/bioinformatics/btaf148}
}
Kostenlos starten →