Dimensionality reduction by UMAP for visualizing and aiding in classification of imaging flow cytometry data

Area of Science:

• Biotechnology
• Computational Biology
• Data Science

Background:

• Imaging flow cytometry (IFC) generates high-dimensional data for single-cell analysis.
• Existing analytical methods for IFC data have limitations in speed, information utilization, and annotation requirements.
• Innovative approaches are needed to effectively analyze complex IFC datasets.

Purpose of the Study:

• To develop and evaluate a novel analytical workflow for imaging flow cytometry data.
• To leverage dimensionality reduction techniques, specifically Uniform Manifold Approximation and Projection (UMAP), for enhanced IFC data analysis.
• To provide a robust and efficient method for visualizing, clustering, and annotating unannotated objects in large-scale IFC experiments.

Main Methods:

• Application of a Uniform Manifold Approximation and Projection (UMAP) workflow to diverse imaging flow cytometry datasets.
• Comparative analysis of UMAP against other popular dimensionality reduction methods.
• Implementation of UMAP for fast visualization, clustering, and tagging of cellular populations.

Main Results:

• The UMAP workflow demonstrated superior speed and accuracy in analyzing IFC datasets compared to existing methods.
• UMAP facilitated rapid visualization and effective clustering of cellular populations.
• The method enabled efficient tagging of unannotated objects within large-scale experiments.

Conclusions:

• The developed UMAP workflow offers a powerful and efficient solution for analyzing complex, high-dimensional imaging flow cytometry data.
• This approach can significantly improve the analysis of single-cell heterogeneity and rare event detection.
• The UMAP workflow is a valuable tool for IFC data analysis, adaptable for use with or as a precursor to deep learning methods.