MetaGXplore: Integrating Multi-Omics Data with Graph Convolutional Networks for Pan-cancer Patient Metastasis Identification

Tao Jiang, Haiyang Jiang, Xinyi Ma, Minghao Xu, Yan Liang, Wentao Zhang

Abstract: Tumor metastasis, the spread of cancer cells to distant sites, significantly challenges clinical prognosis and treatment efficacy, ultimately impacting patient survival. Current predictive methods rely on extensive examinations and subjective clinical judgments, underscoring the urgent need for accurate and rapid assessments of metastasis likelihood to guide effective interventions. Moreover, identifying key genes closely associated with metastasis probability is essential for gaining insights that could facilitate the identification of metastasis-specific biomarkers. We developed MetaGXplore, an innovative Graph Convolutional Neural Network (GCN)-based framework, designed to predict metastasis probability by integrating pan-cancer multi-omic datasets with a protein-protein interaction network while identifying genes crucial to the metastatic process. By leveraging multi-omics datasets, our approach offers a comprehensive view of cancer biology, enhancing accuracy in metastasis forecasting through advanced deep learning algorithms. The efficacy of MetaGXplore was validated through extensive model evaluations, graph structural analysis, and multi-omics data assessment. We applied GNNExplainer to interpret our classification model’s outcomes, providing insights into key feature contributions. Additionally, enrichment analysis of pivotal genes further elucidated the biological mechanisms underlying metastasis.