Network based simultaneous embedding of cells and marker genes from scRNA-seq studies

Area of Science:

• Genomics
• Bioinformatics
• Computational Biology

Background:

• Single-cell RNA sequencing (scRNA-seq) generates complex datasets requiring advanced analytical methods.
• Current clustering and visualization tools often struggle to effectively represent both cellular populations and their associated marker genes simultaneously.

Purpose of the Study:

• To introduce Stardust, an iterative, force-directed graph layout algorithm for scRNA-seq data analysis.
• To enable the simultaneous embedding and visualization of cells and marker genes on a single 2D map.
• To provide a flexible visualization pipeline compatible with existing methods like UMAP and t-SNE.

Main Methods:

• Development of Stardust, an iterative, force-directed graph layout algorithm.
• Implementation of a novel visualization pipeline for simultaneous cell and marker gene embedding.
• Benchmarking Stardust against established visualization and clustering tools using scRNA-seq and spatial transcriptomics datasets.

Main Results:

• Stardust successfully integrates cells and marker genes into a single 2D visualization.
• The algorithm demonstrates competitive performance in identifying and visualizing cell types accurately.
• Spatial coherence of cell type visualization is maintained in transcriptomics datasets.

Conclusions:

• Stardust offers a novel, unified approach for visualizing complex scRNA-seq and spatial transcriptomics data.
• The algorithm enhances the ability to accurately identify and spatially visualize cell types.
• Stardust provides a valuable tool for exploring cellular heterogeneity and gene expression patterns.