Remics: a redescription-based framework for multi-omics analysis.

Area of Science:

• Computational biology
• Genomics
• Systems biology

Background:

• Complex diseases like cancer involve intricate molecular mechanisms across multiple 'omic' layers.
• Integrating and interpreting multi-omics data is crucial for understanding disease subtypes, identifying biomarkers, and improving prognostic models.

Purpose of the Study:

• Introduce Remics, a redescription-based framework for biologically meaningful multi-omics data integration.
• Leverage higher-order statistical representations to uncover cross-omics feature associations and molecular interactions.

Main Methods:

• Remics utilizes higher-order cumulants to identify 'redescriptions'—feature sets capturing equivalent biological variation across modalities.
• Analyzes feature groups via network representations, multi-omics risk scoring, and biomarker discovery.
• Applies the framework to simulated data and The Cancer Genome Atlas multi-omics data from six cancer types.

Main Results:

• Redescription-based integration reveals functionally coherent cross-omics feature associations.
• Demonstrates improved interpretability and discovery of novel molecular relationships compared to state-of-the-art methods.
• Highlights the potential of higher-order multi-omics statistical analysis in advancing precision medicine.

Conclusions:

• Remics provides a powerful framework for interpreting complex multi-omics data.
• Facilitates the discovery of novel molecular insights into disease mechanisms.
• Advances precision medicine through enhanced interpretability and biomarker identification.