A supervised Bayesian factor model for the identification of multi-omics signatures.

Area of Science:

• Biostatistics
• Bioinformatics
• Computational Biology

Background:

• Predictive biological signatures are crucial biomarkers for disease diagnosis, prognosis, and treatment response.
• High-throughput omics data require dimensionality reduction for signature discovery and mechanistic insights.
• Current methods often perform multi-omics integration and predictive modeling separately, leading to suboptimal results.

Approach:

• Developed a supervised variational Bayesian factor model for multi-omics signature extraction.
• Introduced Signature-based multiPle-omics intEgration via lAtent factoRs (SPEAR), an R-package.
• SPEAR adaptively determines factor rank, structure emphasis, data relevance, and feature sparsity.

Key Points:

• SPEAR improves the reconstruction of underlying factors compared to traditional methods.
• Demonstrated enhanced prediction accuracy for COVID-19 severity.
• Showcased improved prediction accuracy for breast cancer tumor subtypes.

Conclusions:

• Integrated multi-omics data analysis and predictive modeling yield more effective biological signatures.
• SPEAR offers a powerful tool for discovering predictive multi-omics signatures.
• The SPEAR R-package is publicly available for broader research application.