Single-cell alternative splicing analysis with Expedition reveals splicing dynamics during neuron differentiation

Alternative splicing (AS) generates isoform diversity critical for cellular identity and homeostasis, yet characterization of this diversity in single cells remains limited. We developed Expedition, a computational framework to categorize and visualize the heterogeneity of AS from single-cell transcriptomes. Expedition consists of (i) outrigger, a de novo splice graph transversal algorithm to detect AS from single cell RNA-seq; (ii) anchor, a Bayesian approach to assign splicing modalities and (iii) bonvoyage, using non-negative matrix factorization to visualize modality changes. By applying Expedition to single iPSCs undergoing neuron differentiation, we discover that 25% of AS exons exhibit bimodality and are flanked by longer and more conserved introns harboring distinct cis-regulatory motifs. Bimodal exons are highly dynamic during cellular transitions, preserve translatability, enriched in recently emerged genes and have conserved AS in mammals. Applying Expedition (http://github.com/YeoLab/Expedition) in single cells redefines our estimates and understanding of AS in evolution and biology. Analysis of alternative splicing changes across motor neuron differentiation