Coverage-dependent bias creates the appearance of binary splicing in single cells.

Area of Science:

• Molecular Biology
• Genomics
• Bioinformatics

Background:

• Single-cell RNA sequencing (scRNA-seq) is a powerful tool for understanding cellular identity.
• Previous studies observed a surprising bimodal pattern in alternative splicing across single cells, suggesting cells consistently produce one of two isoforms.

Purpose of the Study:

• To investigate the cause of the observed bimodal pattern in alternative splicing in single-cell RNA sequencing data.
• To determine if the bimodal splicing pattern is a true biological phenomenon or a technical artifact.

Main Methods:

• Analysis of alternative splicing in human and mouse single-cell RNA sequencing datasets.
• Development and application of a probabilistic simulator to model scRNA-seq data.
• Evaluation of the impact of gene expression levels and capture efficiency on isoform detection.

Main Results:

• Simulations demonstrated that low gene expression and low capture efficiency significantly distort observed isoform distributions.
• The apparent binary splicing pattern arises primarily from technical limitations inherent in scRNA-seq.
• The study identified that the true underlying splicing may not be strictly bimodal.

Conclusions:

• The bimodal pattern in single-cell alternative splicing is largely an artifact of technical limitations, not a biological reality.
• Accurate interpretation of single-cell splicing requires accounting for data sparsity and capture efficiency.
• Developing methods to correct for technical noise can yield more biologically meaningful measurements of alternative splicing in single cells.