Background

RNA alternative splicing (AS) is a fundamental regulatory mechanism in eukaryotic cells that expands transcriptomic and proteomic diversity by generating multiple RNA transcripts and protein isoforms from a single gene through differential exon usage [1]. This process plays critical roles in cell fate specification, tissue development, and the regulation of cellular states, and its dysregulation has been implicated in numerous diseases, including neurological disorders and cancer [2,3]. Understanding AS at single-cell resolution is therefore essential for dissecting cellular heterogeneity and dynamic state transitions in complex biological systems.

Advances in single-cell RNA sequencing (scRNA-seq) technologies have enabled transcriptome-wide profiling of individual cells; however, most widely used platforms, such as droplet-based methods [4,5], are limited in their ability to accurately quantify full-length transcripts and splice isoforms. Full-length scRNA-seq protocols such as SMART-Seq2 [6] enable accurate detection of splice junctions and isoform usage, providing a strong foundation for AS analysis at single-cell resolution [7]. Computational tools such as MARVEL [8] facilitate quantification and downstream analysis of splicing events, including calculation of percent spliced-in (PSI) values, differential AS analysis, and dimensionality reduction methods such as principal component analysis (PCA) and uniform manifold approximation and projection (UMAP). While MARVEL provides a robust framework for AS quantification in single-cell RNA-seq data, it has relatively limited built-in support for further downstream analyses, such as filtering AS events, clustering cells using AS features, and visualization of specific AS events. As a result, constructing a reproducible, end-to-end workflow from raw sequencing data to integrated biological interpretation remains challenging.

This protocol provides a comprehensive and stepwise workflow for analyzing AS using SMART-Seq2 scRNA-seq data, integrating gene expression and splicing analyses within a unified framework. Starting from raw sequencing reads, the pipeline includes read alignment, gene and splice junction quantification, mutation detection, and gene expression–based clustering and annotation. AS analysis is then performed using MARVEL in combination with customized steps to calculate PSI values, identify variable AS events, perform dimensionality reduction, cluster cells based on AS features, and conduct differential splicing analysis. This workflow emphasizes reproducibility, flexibility, and compatibility with standard single-cell analysis pipelines.

While this protocol is demonstrated in the context of pediatric high-grade glioma (pHGG) with H3.3K27M mutation, it is broadly applicable to other biological systems where full-length scRNA-seq data are available. It can be adapted to study AS dynamics in normal development, disease progression, and treatment response, and may be extended to investigate splicing regulation, isoform-specific functions, and interactions with genetic or epigenetic alterations.