U-CIE [/juː 'siː/]: Color encoding of high-dimensional data.

Area of Science:

• Computational biology
• Data visualization
• Bioinformatics

Background:

• High-dimensional data, especially from omics studies, presents visualization challenges.
• Existing methods struggle to jointly display multiple data types like single-cell expression and gene networks.
• Effective visualization is crucial for identifying patterns and anomalies in large datasets.

Purpose of the Study:

• To introduce U-CIE, a novel visualization method for high-dimensional data.
• To enable the joint visualization of diverse data types while preserving original data structure.
• To demonstrate the broad applicability of U-CIE across different scientific domains.

Main Methods:

• U-CIE combines Uniform Manifold Approximation and Projection (UMAP) for dimensionality reduction with the CIELAB color space.
• UMAP reduces data to three dimensions, preserving local and global data structure.
• The 3D data is then mapped to the perceptually uniform CIELAB color space for color encoding.

Main Results:

• U-CIE successfully encodes high-dimensional data into colors, retaining significant structural information.
• The method demonstrated effective visualization of single-cell data integrated with protein networks.
• Applications to metagenomic data visualization on maps and scatter plots were also successful.

Conclusions:

• U-CIE offers a powerful approach for visualizing complex, high-dimensional datasets, particularly omics data.
• The combination of UMAP and CIELAB provides a perceptually meaningful color encoding that reflects data structure.
• U-CIE is a versatile tool applicable to various biological and scientific data visualization challenges.