Visualization and cellular hierarchy inference of single-cell data using SPADE.

Area of Science:

• Computational Biology
• Bioinformatics
• Data Science

Background:

• High-throughput single-cell technologies reveal cellular heterogeneity but present analysis challenges.
• Existing methods for single-cell data visualization and hierarchy inference have limitations.

Purpose of the Study:

• To present Spanning-tree Progression Analysis of Density-normalized Events (SPADE) as a robust tool for single-cell data analysis.
• To demonstrate SPADE's applicability to diverse single-cell datasets, including RNA-seq.
• To compare SPADE with t-distribution stochastic neighborhood embedding (t-SNE) and propose an integrated approach.

Main Methods:

• Implementation of SPADE using an open-source R package compatible with major operating systems.
• Application of SPADE to mass cytometry data of hematopoietic cells.
• Comparative analysis of SPADE and t-SNE for visualization and hierarchy inference.

Main Results:

• SPADE provides efficient density-based visualization and cellular hierarchy inference.
• SPADE analysis is rapid, completing in approximately 5 minutes on standard hardware.
• SPADE demonstrates applicability to single-cell RNA-seq data, complementing its use in cytometry.
• An integrative strategy combining SPADE and t-SNE enhances cellular hierarchy inference.

Conclusions:

• SPADE is a versatile and efficient algorithm for single-cell data analysis and visualization.
• SPADE offers a valuable approach for inferring cellular hierarchies from complex single-cell datasets.
• Combining SPADE with t-SNE provides a powerful integrated strategy for advanced single-cell data interpretation.