Multi-Omics Data Integration Clustering for Cancer Subtypes Identification Based on Motif High-Order Similarity Graph...

Area of Science:

• Bioinformatics
• Computational Biology
• Cancer Research

Background:

• Accurate cancer subtype identification is crucial and increasingly relies on multi-omics data integration.
• Existing similarity graph methods often overlook high-order sample relationships and complementary omics information.

Purpose of the Study:

• To develop an innovative multi-omics integration framework, MHSGTR, for enhanced cancer subtype identification.
• To address limitations of existing methods by incorporating high-order neighborhood information and exploring omics complementarity.

Main Methods:

• MHSGTR utilizes Motif theory to construct high-order similarity graphs, capturing latent structural information.
• Third-order tensor regularization and an attention module are employed for multi-omics data integration and consensus graph construction.
• Spectral clustering is applied for final cancer subtype identification.

Main Results:

• MHSGTR demonstrated superior clustering performance across eight TCGA cancer datasets.
• A case study on adrenocortical carcinoma (ACC) revealed biologically significant cancer subtypes.
• The framework effectively integrates multi-omics data, showcasing robust performance.

Conclusions:

• MHSGTR offers an effective approach for cancer subtype discovery through advanced multi-omics integration.
• The method's ability to leverage high-order relationships and omics complementarity enhances biological insights.
• This framework holds promise for advancing precision oncology.