MOUSSE: Multi-Omics Using Subject-Specific SignaturEs.

Area of Science:

• Bioinformatics
• Computational Biology
• Genomics

Background:

• High-throughput technologies generate vast multi-omics data (genomics, proteomics, metabolomics, transcriptomics).
• Individual omics analysis is insufficient for understanding complex diseases like cancer.
• A lack of standardized methods hinders multi-omics data integration.

Purpose of the Study:

• Introduce MOUSSE, a novel normalization-free pipeline for unsupervised multi-omics integration.
• Develop a method to effectively merge similarity networks derived from individual omics data.
• Analyze cancer patient survival and identify key molecular features.

Main Methods:

• Utilized rank-based subject-specific signatures for unsupervised integration.
• Constructed subject similarity networks for each omics layer.
• Merged individual similarity networks based on informative content.
• Applied the pipeline to survival data from ten cancer types.

Main Results:

• Achieved meaningful patient clustering across different cancer types.
• Demonstrated superior average classification performance compared to ten state-of-the-art algorithms.
• Identified biologically relevant features from subject-specific signatures linked to cancer progression and survival.

Conclusions:

• MOUSSE provides an effective approach for unsupervised multi-omics integration.
• The pipeline accurately clusters patients and predicts survival outcomes.
• Identified molecular features offer insights into cancer progression and differential survival.