shinyUMAP: an online tool for promoting understanding of single cell omics data visualization

Area of Science:

• Computational Biology
• Bioinformatics
• Data Visualization

Background:

• High-dimensional single-cell omics data, including single-cell RNA sequencing (scRNA-seq), heavily relies on visualization techniques for interpretation.
• Uniform Manifold Approximation and Projection (UMAP) is a prevalent dimensionality reduction technique used in single-cell omics research.
• Concerns exist regarding UMAP's ability to accurately represent global cell-cell relationships and spatial distances between cell clusters in 2D projections.

Purpose of the Study:

• To address the potential misinterpretation of inter-cluster relationships in single-cell studies due to UMAP limitations.
• To provide researchers with a tool to interactively explore how different UMAP hyperparameters affect data visualization.
• To promote the appropriate use of UMAP in single-cell omics data analysis.

Main Methods:

• Development of an interactive online server for single-cell data analysis.
• Implementation of UMAP with adjustable hyperparameters.
• User-driven exploration of UMAP projections with varying parameters.

Main Results:

• The server allows users to upload their single-cell data and generate UMAP visualizations.
• Users can observe how changes in UMAP hyperparameters influence the spatial arrangement of cell clusters.
• The interactive nature of the server facilitates a deeper understanding of UMAP's behavior and limitations.

Conclusions:

• The developed online server aids researchers in appreciating the nuances of UMAP visualization in single-cell omics.
• It serves as an educational tool to prevent common pitfalls associated with misinterpreting UMAP-generated cell cluster relationships.
• Promoting proper usage of UMAP enhances the reliability of single-cell omics data interpretation.