Comparison of methods to detect differentially expressed genes between single-cell populations

Area of Science:

• Genomics
• Bioinformatics
• Computational Biology

Background:

• Single-cell RNA sequencing (scRNA-seq) enables gene expression analysis at the individual cell level.
• Detecting differentially expressed genes (DEGs) is crucial for understanding cell type distinctions.
• The applicability of bulk RNA-seq DEG methods to scRNA-seq data remains an open question.

Purpose of the Study:

• To compare the performance of five statistical methods for DEG detection in scRNA-seq data.
• To evaluate whether existing bulk RNA-seq DEG methods are suitable for scRNA-seq.
• To identify robust DEG detection methods for single-cell population comparisons.

Main Methods:

• Comparative analysis of five statistical DEG detection methods.
• Evaluation across three distinct single-cell comparison settings.
• Assessment of detection counts, sensitivity, and specificity.

Main Results:

• Significant variations were observed in DEG detection numbers, sensitivity, and specificity among the tested methods.
• No systematic advantages were found for existing single-cell-specific DEG methods.
• A previously developed reproducibility-optimization method demonstrated consistent good performance without scRNA-seq specific adjustments.

Conclusions:

• The choice of DEG detection method significantly impacts scRNA-seq analysis outcomes.
• Existing single-cell specific methods do not consistently outperform general approaches.
• A reproducibility-optimization strategy offers a robust and adaptable method for DEG detection in scRNA-seq data analysis.