Search

Enhancing Protein-Protein Interaction Prediction with Graph Neural Networks

Author(s)

Fanjin Zeng*

Affiliation(s)

Xi’an Jiaotong-Liverpool University, Suzhou, China *Corresponding Author

Abstract

Protein-protein interactions (PPIs) are crucial for understanding the complex biochemical networks within living organisms and play significant roles in processes such as cellular signaling, molecular transport, and metabolism. Predicting PPIs is essential for advancing medical research, drug discovery, and disease treatment. Traditional methods, including molecular docking and sequence alignment, have been widely used, but they are often limited by scalability issues, high computational costs, and the difficulty of handling large datasets. Recently, Graph Neural Networks (GNNs) have emerged as a powerful tool for predicting PPIs due to their ability to efficiently process graph-structured data and capture intricate relationships between proteins. This paper explores the application of GNNs in PPI prediction, emphasizing their advantages over traditional approaches, particularly in terms of prediction accuracy, scalability, and efficiency. A novel model leveraging GNNs is presented, demonstrating significant improvements in the prediction of protein interactions, even in large-scale networks. The study shows that GNN-based models can handle omplex, multidimensional data and predict previously unknown interactions with high accuracy, making them a promising tool for drug discovery and disease modeling.

Keywords

Protein-Protein Interaction (PPI); Graph Neural Networks (GNN); Machine Learning

References

[1] Poluri K M, Gulati K, Sarkar S. Protein-protein interactions[M]. Springer, 2021. [2] Xiang H, Zhou M, Li Y, et al. Drug discovery by targeting the protein–protein interactions involved in autophagy[J]. Acta Pharmaceutica Sinica B, 2023, 13(11): 4373-4390. [3] Hasan M R, Paul B K, Ahmed K, et al. Design protein-protein interaction network and protein-drug interaction network for common cancer diseases: A bioinformatics approach[J]. Informatics in Medicine Unlocked, 2020, 18: 100311. [4] Hashemi Z S, Zarei M, Fath M K, et al. In silico approaches for the design and optimization of interfering peptides against protein–protein interactions[J]. Frontiers in Molecular Biosciences, 2021, 8: 669431. [5] Wang L, Li F, Ma X, et al. PPI-Miner: a structure and sequence motif co-driven protein–protein interaction mining and modeling computational method[J]. Journal of Chemical Information and Modeling, 2022, 62(23): 6160-6171. [6] Longa A, Azzolin S, Santin G, et al. Explaining the explainers in graph neural networks: a comparative study[J]. ACM Computing Surveys, 2025, 57(5): 1-37. [7] Kiouri D P, Batsis G C, Chasapis C T. Structure-Based Approaches for Protein–Protein Interaction Prediction Using Machine Learning and Deep Learning[J]. Biomolecules, 2025, 15(1): 141. [8] Meng L, Wei L, Wu R. MVGNN-PPIS: A novel multi-view graph neural network for protein-protein interaction sites prediction based on Alphafold3-predicted structures and transfer learning[J]. International Journal of Biological Macromolecules, 2025: 140096. [9] Zhang Z, Zhang Q, Xiao J, et al. MFC-PPI: protein–protein interaction prediction with multimodal feature fusion and contrastive learning[J]. The Journal of Supercomputing, 2025, 81(4): 1-19. [10] Jin M, Xue H, Wang Z, et al. ProLLM: protein chain-of-thoughts enhanced LLM for protein-protein interaction prediction[J]. bioRxiv, 2024: 2024.04. 18.590025.