MORE: a multi-omics data-driven hypergraph integration network for biomedical data classification and biomarker...

Area of Science:

• Biomedical Informatics
• Genomics
• Computational Biology

Background:

• High-throughput sequencing has revolutionized omics research, enabling complex biological studies.
• Integrating multi-omics data offers potential for deeper insights into biological mechanisms.
• Existing graph-based methods often fail to capture cross-modal correlations and treat modalities equally.

Purpose of the Study:

• To develop a novel approach for integrating multi-omics data that addresses limitations of existing methods.
• To improve the understanding of complex diseases by leveraging complementary information across omics modalities.
• To enhance biomarker discovery for diseases like Alzheimer's, breast cancer, and glioblastoma.

Main Methods:

• Introduced Multi-Omics hypeRgraph integration nEtwork (MORE), a novel approach for multi-omics data integration.
• Constructed comprehensive hyperedge groups to capture intra- and inter-modal correlations.
• Employed multi-omics hypergraph encoding and self-attention mechanisms for representation learning and prediction.

Main Results:

• MORE demonstrated superior performance in classification tasks for Alzheimer's disease, invasive breast carcinoma, and glioblastoma.
• The method effectively identified a wider range of disease-related biomarkers compared to state-of-the-art approaches.
• Results highlight MORE's advantage in biomedical data mining and interpretation.

Conclusions:

• MORE is a valuable tool for facilitating multi-omics analysis and novel biomarker discovery.
• The approach enhances the understanding of complex diseases by integrating diverse omics data.
• The study provides a competitive method for advancing biomedical research through data integration.