Detecting differential alternative splicing events in scRNA-seq with or without Unique Molecular Identifiers

Area of Science:

• Genomics
• Bioinformatics
• Molecular Biology

Background:

• Single-cell RNA sequencing (scRNA-seq) enables gene expression analysis at the individual cell level.
• scRNA-seq data presents challenges due to high noise, technical variability, and low sequencing depth.
• Investigating splicing heterogeneity is crucial but difficult with current scRNA-seq methods.

Purpose of the Study:

• To develop a sensitive method for differential splicing analysis in scRNA-seq data.
• To address the challenges of noise and low coverage in scRNA-seq.
• To improve the detection of splicing variations across individual cells.

Main Methods:

• Proposed SCATS (Single-Cell Analysis of Transcript Splicing) for differential splicing analysis.
• SCATS models scRNA-seq data (UMI and non-UMI) accounting for technical noise, capture efficiency, amplification bias, and transcriptional burstiness.
• Key innovation: grouping exons from the same isoforms to aggregate reads and enhance detection at low sequencing depth.

Main Results:

• SCATS demonstrated a well-controlled type I error rate and higher statistical power compared to existing methods (Census, DEXSeq), especially for subtle splicing differences.
• Census showed type I error inflation, while DEXSeq was more conservative.
• Applied to mouse brain scRNA-seq data, SCATS identified more differential splicing events than Census and DEXSeq.

Conclusions:

• SCATS is a sensitive and robust tool for differential splicing analysis in scRNA-seq data.
• The method effectively handles technical noise and low sequencing depth, improving detection of subtle splicing variations.
• SCATS is well-suited for various scRNA-seq splicing studies and is available at https://github.com/huyustats/SCATS.