Stabilized marker gene identification and functional annotation from single-cell transcriptomic data.

Area of Science:

• Computational Biology
• Genomics
• Bioinformatics

Background:

• Single-cell transcriptomics generates large datasets, necessitating reproducible marker gene identification and functional cell annotation.
• Conventional differential gene expression (DEG) analysis methods lack consistency across datasets and functional insights.
• Accurate cell type identification is crucial for understanding cellular heterogeneity and function.

Purpose of the Study:

• To introduce scSCOPE, an R-based platform for identifying reproducible marker genes and associated pathways in single-cell RNA sequencing (scRNAseq) data.
• To improve cell type annotation and facilitate functional investigations of cellular heterogeneity.
• To provide a robust alternative to conventional DEG analysis for scRNAseq data.

Main Methods:

• Utilizes stabilized LASSO (Least Absolute Shrinkage and Selection Operator) for feature selection.
• Employs bootstrapped co-expression networks to identify gene relationships.
• Integrates pathway enrichment analysis for functional annotation of marker genes.

Main Results:

• scSCOPE identified reproducible, cell type-specific marker genes and pathways with high consistency across 9 diverse scRNAseq datasets (human and mouse immune cells).
• Demonstrated superior performance compared to conventional methods in marker gene identification and functional annotation.
• Provided in-depth molecular insights into the functionality of identified marker genes through co-expression and pathway analyses.

Conclusions:

• scSCOPE offers a robust and consistent approach for marker gene discovery and functional annotation in scRNAseq data.
• The platform enhances the accuracy of cell type annotation and accelerates experimental validation.
• scSCOPE is expected to significantly advance research on cellular heterogeneity and function.