GAUDI: interpretable multi-omics integration with UMAP embeddings and density-based clustering.

Area of Science:

• Computational Biology
• Bioinformatics
• Systems Biology

Background:

• Integrating multi-omics data is essential for a comprehensive understanding of cellular control mechanisms.
• Single 'omic approaches often struggle to capture complex inter-omic relationships.
• Existing methods may not fully capture non-linear interactions within high-dimensional biological datasets.

Purpose of the Study:

• To develop a novel, unsupervised method for integrating high-dimensional multi-omics data.
• To uncover complex, non-linear relationships between different omics layers.
• To provide interpretable visualizations for identifying biological insights and potential biomarkers.

Main Methods:

• Developed GAUDI (Group Aggregation via UMAP Data Integration), a non-linear, unsupervised integration method.
• Utilized independent UMAP embeddings for concurrent analysis of multiple omics datasets.
• Employed sample clustering based on multi-omic profiles and identification of latent factors.

Main Results:

• GAUDI effectively uncovers non-linear relationships among multi-omics data, outperforming state-of-the-art methods.
• The method successfully clusters samples based on integrated multi-omic profiles.
• Identified latent factors within each omics dataset, facilitating interpretation of cluster-driving features.

Conclusions:

• GAUDI offers a powerful approach for integrating and analyzing multi-omics data.
• The method enhances interpretability and visualization, aiding in the discovery of novel biological insights.
• GAUDI facilitates the identification of potential biomarkers across diverse experimental designs.