Sparse multiple co-Inertia analysis with application to integrative analysis of multi -Omics data.

Area of Science:

• Bioinformatics
• Computational Biology
• Systems Biology

Background:

• Multiple co-inertia analysis (mCIA) integrates multiple datasets but yields non-sparse results, hindering feature identification.
• Interpreting complex relationships in high-dimensional omics data remains a challenge.

Purpose of the Study:

• Introduce novel sparse and structured sparse mCIA methods.
• Enhance feature selection and interpretability in multi-omics data analysis.

Main Methods:

• Developed sparse mCIA for direct sparse loading estimation.
• Created structured sparse mCIA to incorporate prior biological network information.

Main Results:

• Simulations show sparse and structured sparse mCIA outperform existing mCIA in feature selection and accuracy.
• Applied methods to transcriptomics and proteomics data, identifying cancer-related biomarkers.

Conclusions:

• Sparse mCIA enables simultaneous model estimation and feature selection for improved interpretability.
• Structured sparse mCIA effectively leverages network information, enhancing feature selection and interpretability in omics integration.