Decoding Multi-Omics Signatures in Lower-Grade Glioma Using Protein-Protein Interaction-Informed Graph Attention Networks and Ensemble Learning

Abstract

Background/Objectives: Lower-grade gliomas (LGGs) are a biologically and clinically heterogeneous group of brain tumors, for which molecular stratification plays essential role in diagnosis, prognosis, and therapeutic decision-making. Conventional unimodal classifiers do not necessarily describe cross-layer regulatory dynamics which entail the heterogeneity of glioma. Methods: This paper presents a protein-protein interaction (PPI)-informed hybrid model that combines multi-omics profiles, including RNA expression, DNA methylation, and microRNA expression, with a Graph Attention Network (GAT), Random Forest (RF), and logistic stacking ensemble learning. The proposed model utilizes ElasticNet-based feature selection to obtain the most informative biomarkers across omics layers, and the GAT module learns the biologically significant topological representations in the PPI network. The Synthetic Minority Over-Sampling Technique (SMOTE) was used to mitigate the class imbalance, and the model performance was assessed using a repeated five-fold stratified cross-validation approach using the following performance metrics: accuracy, precision, recall, F1-score, ROC-AUC, and AUPRC. Results: The findings illustrate that a combination of multi-omics data increases subtype classification rates (up to 0.984 ± 0.012) more than single-omics methods, and DNA methylation proves to be the most discriminative modality. In addition, analysis of interpretability using attention revealed the major subtype-specific biomarkers, including UBA2, LRRC41, ANKRD53, and WDR77, that show great biological relevance and could be used as diagnostic and therapeutic tools. Conclusions: The proposed multi-omics based on a biological and explainable framework provides a solid computational approach to molecular stratification and biomarker identification in lower-grade glioma, bridging between predictive power, biological clarification, and clinical benefits.

Keywords: biomarker discovery; graph attention network; lower-grade glioma; multi-omics integration; protein–protein interaction network.

Figure 1

The hybrid multi-omics classification framework integrating ElasticNet feature selection, RF feature importance, and GAT-based graph learning with logistic regression stacking for biomarker discovery and class prediction.

Figure 2

Comprehensive intra- and inter-omics correlation summary across RNA-Seq, DNA methylation, and miRNA layers (Top 300 features per layer).

Figure 3

Precision–recall and ROC curves for LGG subtype classification using the RF-GAT hybrid model.

Figure 4

Integrative multi-omics clustering of lower-grade glioma samples based on the top-500 features identified by the hybrid RF-GAT model.

Figure 5

Top-20 candidate biomarkers contributing to LGG subtype discrimination identified by the RF+GAT hybrid framework.

Figure 6

Differential expression of the top 10 biomarkers across LGG subtypes (normalized 0–1 per gene).

Figure 7

GO and KEGG enrichment of the top 20 biomarkers identified by the RF-GAT model. Bubble plot illustrates the top enriched GO biological processes (blue) and KEGG pathways (orange) based on −log₁₀(adjusted p-value). The bubble size represents the number of genes contributing to each term. The 24 instances of the genes in the figure indicate overlapping genes that play a significant role in various terms.