Unsupervised Multi-Omics Data Integration Methods: A Comprehensive Review.

Area of Science:

• Bioinformatics
• Computational Biology
• Genomics

Background:

• Advancements in Omics technologies generate large-scale datasets (e.g., TCGA, ADNI, GTEx).
• Holistic study of complex biological processes and disease mechanisms requires integrating diverse data modalities.
• Leveraging external biological knowledge from databases is essential for comprehensive analysis.

Purpose of the Study:

• To provide an overview of multi-Omics data integration methods.
• To focus on unsupervised learning approaches for biological data analysis.
• To cover critical components for data integration, including feature selection and available resources.

Main Methods:

• Review of statistical approaches for multi-Omics data integration.
• Focus on unsupervised learning tasks: disease onset prediction, biomarker discovery, disease subtyping, module discovery, and network/pathway analysis.
• Brief review of feature selection methods and relevant datasets.

Main Results:

• Identified various statistical methods applicable to multi-Omics integration.
• Highlighted the utility of unsupervised learning for key biological questions.
• Cataloged essential resources and tools for implementing these methods.

Conclusions:

• Multi-Omics data integration is vital for a comprehensive understanding of biological complexity.
• Unsupervised learning offers powerful tools for extracting insights from integrated Omics data.
• Availability of datasets, feature selection techniques, and computational tools facilitates multi-Omics research.