Application of Aligned-UMAP to longitudinal biomedical studies

Area of Science:

• Computational Biology
• Data Science
• Bioinformatics

Background:

• High-dimensional data analysis is crucial for understanding complex biological systems.
• Existing dimensionality reduction methods are often limited to cross-sectional data.
• Visualizing longitudinal, high-dimensional datasets presents a significant challenge.

Purpose of the Study:

• To introduce and evaluate Aligned-UMAP for visualizing high-dimensional longitudinal data.
• To demonstrate the utility of Aligned-UMAP in identifying patterns and trajectories in biological data.
• To provide guidance on parameter tuning for optimal performance of Aligned-UMAP.

Main Methods:

• Utilized the Aligned-UMAP algorithm, an extension of uniform manifold approximation and projection (UMAP).
• Benchmarked the algorithm's performance on high-dimensional longitudinal datasets.
• Made the implementation code open source to ensure reproducibility and broad applicability.

Main Results:

• Aligned-UMAP effectively visualizes high-dimensional longitudinal datasets.
• The algorithm enables identification of significant patterns and trajectories in biological data.
• Algorithm parameter selection critically impacts the quality of visualization and analysis.

Conclusions:

• Aligned-UMAP is a valuable tool for exploring complex, high-dimensional longitudinal biomedical data.
• Careful tuning of Aligned-UMAP parameters is necessary to unlock its full potential.
• Open-source availability enhances the adoption and advancement of Aligned-UMAP in research.