Deciphering splicing heterogeneity at single-cell resolution by SCSES

TL;DR

SCSES is a new tool that improves the analysis of alternative splicing in single-cell RNA sequencing data, revealing hidden splicing patterns and cell subgroups.

Contribution

SCSES introduces a novel computational framework for estimating splicing heterogeneity by sharing information across similar cells and events.

Findings

SCSES outperforms existing methods in recovering splicing changes and diversity in simulated data.

SCSES identifies splicing heterogeneity and cell subgroups with unique splicing patterns not detected by standard clustering.

The tool is versatile and applicable across different species and sequencing platforms.

Abstract

Alternative splicing (AS) plays a critical role in generating cellular transcriptomic heterogeneity. While single-cell RNA sequencing (scRNA-seq) has become a standard approach for exploring this heterogeneity, it remains challenging to accurately characterize splicing changes at the single-cell level due to high dropout rates, inevitable noise, and limited coverage. To address this, we developed SCSES (Single-Cell Splicing EStimation), a computational framework designed to enhance the AS profiles. SCSES infers and completes the missing splicing changes by sharing information across similar cells and events with data diffusion. Through systematic simulation studies, SCSES outperforms existing algorithms in recovering percent spliced-in (PSI) values and diversity across cell populations. When applied to various datasets, SCSES uncovers substantial splicing heterogeneity and cell subgroups with exclusive splicing patterns, which cannot be captured by conventional single-cell gene expression clustering. Together, our study provides SCSES as a valuable tool in deciphering splicing heterogeneity and is widely capable of handling different biological scenarios, species and sequencing platforms.

Alternative splicing (AS) is a source of transcriptomic heterogeneity, but analysing splicing changes at single-cell level is challenging due to dropout, noise, and limited coverage in scRNA-seq. Here, the authors propose SCSES, a tool to identify AS events and estimate splicing intensity at single-cell resolution. SCSES reveals high splicing heterogeneity beyond conventional expression-based clustering.

Figures