Comparative Analysis of Multi-Omics Integration Using Graph Neural Networks for Cancer Classification

Authors: Fadi Alharbi, Aleksandar Vakanski, Boyu Zhang, Murtada K. Elbashir, Mohanad Mohammed

This study evaluates graph neural network architectures for multi-omics (MO) data integration based on graph-convolutional networks (GCN), graph-attention networks (GAT), and graph-transformer networks (GTN). Differential gene expression and LASSO (Least Absolute Shrinkage and Selection Operator) regression are employed for reducing the omics data dimensionality and feature selection; hence, the developed models are referred to as LASSO-MOGCN, LASSO-MOGAT, and LASSO-MOGTN. Graph structures constructed using sample correlation matrices and protein-protein interaction networks are investigated. Experimental validation is performed with a dataset of 8,464 samples from 31 cancer types and normal tissue, comprising messenger-RNA, micro-RNA, and DNA methylation data. The results show that the models integrating multi-omics data outperformed the models trained on single omics data, where LASSO-MOGAT achieved the best overall performance, with an accuracy of 95.9%. The findings also suggest that correlation-based graph structures enhance the models’ ability to identify shared cancer-specific signatures across patients in comparison to protein-protein interaction networks-based graph structures.

📁 Repository Organization

The code in the repository presents a comparative analysis for three different types of graph-based neural networks with multi-omics data integration (mRNA, miRNA, and DNA methylation) for cancer prediction.

The architectures employ two types of graph structures: PPI (protein-protein interaction) network and correlation matrix.

The codes that employ PPI graph structure include:

• GAT_MultiOmics_PPI.ipynb: Graph Attention Network (GAT).
• GCN_MultiOmics_PPI.ipynb: Graph Convolutional Network (GCN).
• GTN_MultiOmics_PPI.ipynb: Graph Transformer Network (GTN).

Similarly, the codes that employ correlation matric graph structure include:

• GAT_MultiOmics_CorrMat.ipynb: Graph Attention Network (GAT).
• GCN_MultiOmics_CorrMat.ipynb: Graph Convolutional Network (GCN).
• GTN_MultiOmics_CorrMat.ipynb: Graph Transformer Network (GTN).

📊 Data

The provided file PPI.csv contains information about the PPI (protein-protein interaction) network.

The file containing the multi-omics data (mRNA, miRNA, and DNA methylation) can be downloaded from: here.

▶️ Use

The codes are provided as Jupyter Notebook files. To reproduce the results, run the .ipynb files.

📖 Citation

If you use the codes or the methods in your work, please cite the following article:

@ARTICLE{Alharbi2025,
TITLE={Comparative Analysis of Multi-Omics Integration Using Graph Neural Networks for Cancer Classification},
AUTHOR={Alharbi, Fadi and Vakanski, Aleksandar and Zhang, Boyu and Elbashir, Murtada K. and Mohammed, Mohanad},
JOURNAL = {IEEE Access},
YEAR = {2025},
VOLUME = {13},
PAGES = {37724-37736},
URL = {https://ieeexplore.ieee.org/abstract/document/10879447},
DOI = {10.1109/ACCESS.2025.3540769}
}

🚩 License

MIT License

👏 Acknowledgments

This work was supported by the National Institute of General Medical Sciences of the National Institutes of Health under Award P20GM104420.

✉️ Contact or Questions

A. Vakanski, e-mail: vakanski at uidaho.edu