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

Multi-omics data is increasingly being utilized to advance computationalmethods for cancer classification. However, multi-omics data integration posessignificant challenges due to the high dimensionality, data complexity, anddistinct characteristics of various omics types. This study addresses thesechallenges and evaluates three graph neural network architectures formulti-omics (MO) integration based on graph-convolutional networks (GCN),graph-attention networks (GAT), and graph-transformer networks (GTN) forclassifying 31 cancer types and normal tissues. To address thehigh-dimensionality of multi-omics data, we employed LASSO (Least AbsoluteShrinkage and Selection Operator) regression for feature selection, leading tothe creation of LASSO-MOGCN, LASSO-MOGAT, and LASSO-MOTGN models. Graphstructures for the networks were constructed using gene correlation matricesand protein-protein interaction networks for multi-omics integration ofmessenger-RNA, micro-RNA, and DNA methylation data. Such data integrationenables the networks to dynamically focus on important relationships betweenbiological entities, improving both model performance and interpretability.Among the models, LASSO-MOGAT with a correlation-based graph structure achievedstate-of-the-art accuracy (95.9%) and outperformed the LASSO-MOGCN andLASSO-MOTGN models in terms of precision, recall, and F1-score. Our findingsdemonstrate that integrating multi-omics data in graph-based architecturesenhances cancer classification performance by uncovering distinct molecularpatterns that contribute to a better understanding of cancer biology andpotential biomarkers for disease progression.