From gene correlations to cell clusters: COTAN improved scRNA-seq analysis.

Area of Science:

• Genomics
• Bioinformatics
• Computational Biology

Background:

• Single-cell RNA sequencing (scRNA-seq) generates high-dimensional sparse data.
• Existing analysis pipelines often struggle with sparsity and low-expression genes in droplet-based scRNA-seq.
• Current methods may distort true-zero counts and discard biologically relevant low-abundance regulators.

Purpose of the Study:

• To present an extended COTAN workflow tailored for UMI droplet scRNA-seq data.
• To address challenges of sparsity and low-expression gene detection in scRNA-seq analysis.
• To provide a robust, end-to-end solution for scRNA-seq data analysis.

Main Methods:

• Developed an extended COTAN workflow based on gene correlations.
• The method natively models zero counts without imputation, log-normalization, or scaling.
• Validated on diverse public scRNA-seq datasets.

Main Results:

• Generated robust gene-gene correlation matrices.
• Demonstrated superior detection of differentially expressed genes, including low-expression ones.
• Achieved highly sensitive statistical scoring for detecting cell heterogeneity and uniformity.

Conclusions:

• The extended COTAN workflow offers a biologically grounded solution for scRNA-seq analysis.
• COTAN outperforms leading tools in gene detection and cell group analysis.
• The workflow is suitable for UMI droplet data, handling sparsity effectively.