Algorithms and tools for data-driven omics integration to achieve multilayer biological insights: a narrative review

Area of Science:

• Systems biology
• Bioinformatics
• Computational biology

Background:

• High-throughput omics technologies generate vast datasets crucial for understanding complex biological systems.
• Integrating diverse omics data (genomics, transcriptomics, proteomics, metabolomics) is key to unraveling disease mechanisms and identifying biomarkers.
• Recent literature (2018-2024) highlights evolving strategies for multi-omics data integration.

Purpose of the Study:

• To review and categorize strategies for integrating genomics, transcriptomics, proteomics, and metabolomics data.
• To assess the prevalence and effectiveness of different integration approaches.
• To identify challenges and open issues in multi-omics integration.

Main Methods:

• Literature review focusing on multi-omics integration strategies published between 2018 and 2024.
• Categorization of integration approaches into statistical-based, multivariate, and machine learning/artificial intelligence techniques.
• Analysis of the application and performance of different integration methods.

Main Results:

• Statistical approaches, particularly correlation-based methods, were most prevalent, followed by multivariate and machine learning techniques.
• Multi-omics integration generally enhances the discovery of molecular mechanisms, biomarkers, and classification accuracy compared to single-omics analyses.
• Common challenges include data quality, missing values, collinearity, dimensionality, and data heterogeneity.

Conclusions:

• Multi-omics integration holds significant potential for advancing systems biology and precision medicine.
• Addressing data quality and complexity issues is crucial for unlocking the full benefits of integrating diverse biological datasets.
• Further research is needed to develop robust methods for overcoming remaining challenges in omics data integration.