scDEcrypter: Uncertainty-aware differential expression analysis for viral infection in scRNA-seq

Area of Science:

• Computational biology
• Virology
• Genomics

Background:

• Single-cell RNA sequencing (scRNA-seq) faces challenges in viral infection studies.
• Sparse viral reads and under-labeled cells complicate differential expression (DE) analysis.
• Bystander cell responses can confound results.

Purpose of the Study:

• To introduce scDEcrypter, a new computational tool for analyzing viral infections in scRNA-seq data.
• To improve the identification of infected cell states and associated gene expression.
• To enable more accurate DE analysis in the presence of confounding factors.

Main Methods:

• Developed scDEcrypter, a penalized two-way mixture model.
• Incorporated partial labels for infection status and cell type.
• Utilized data-splitting for robust inference and likelihood-based DE analysis.

Main Results:

• scDEcrypter demonstrated improved recovery of infected cell states in simulations and real data.
• Identified more biologically coherent infection-associated genes.
• Revealed enriched pathways related to viral infection responses.

Conclusions:

• scDEcrypter offers a powerful solution for analyzing viral infections using scRNA-seq.
• The method enhances the accuracy and biological relevance of DE analysis.
• Facilitates a deeper understanding of host-pathogen interactions at the single-cell level.