DECENT: differential expression with capture efficiency adjustmeNT for single-cell RNA-seq data.

Area of Science:

• Genomics
• Bioinformatics
• Computational Biology

Background:

• Dropout events are prevalent in single-cell RNA sequencing (scRNA-seq) data.
• Unaddressed dropout can compromise the statistical validity of analyses, particularly differential expression (DE).
• Existing DE methods often fail to explicitly model the molecular capture process leading to dropout.

Purpose of the Study:

• To develop DECENT, a novel method for DE analysis in scRNA-seq data.
• To explicitly and accurately model the molecule capture process inherent in scRNA-seq experiments.
• To improve the performance of DE analysis by addressing dropout events.

Main Methods:

• Development of DECENT, a statistical method for scRNA-seq data.
• Explicit modeling of the molecule capture process to account for dropout.
• Validation across multiple public scRNA-seq datasets from diverse technological platforms.

Main Results:

• DECENT demonstrates superior DE performance compared to methods that do not model dropout.
• Improvements are consistent across various scRNA-seq datasets and platforms.
• DECENT shows significant gains when the capture process is overdispersed, maintaining type I error control with enhanced sensitivity.
• Performance is robust even without spike-ins for capture model calibration.

Conclusions:

• DECENT offers a more accurate approach to differential expression analysis in scRNA-seq.
• The method effectively addresses the challenge of dropout events, improving analytical validity.
• DECENT is available as a publicly accessible R package.