Abstract
Background High-throughput single-cell RNA-seq (scRNA-seq) is a powerful technology for studying gene expression variability in single cells; however, standard analysis approaches only consider the overall expression of each gene, masking additional heterogeneity that could exist through cell type-specific expression of alternative mRNA transcripts.
Results Here we show that differential transcript usage (DTU) can be readily detected in data-sets generated from commonly used polyA-captured nanodroplet scRNA-seq technology. Our computational pipeline, Sierra, detects and quantifies polyadenylation sites in scRNA-seq data-sets, which are utilised to evaluate DTU between single-cell populations. We validate our approach by comparing cardiac cell populations derived from scRNA-seq to bulk RNA-seq of matched populations obtained through fluorescent activated cell sorting (FACS), demonstrating that we detect a significant overlap in cell-type DTU between the congruous data-sets. We further illustrate the utility of our method by detecting alternative transcript usage in human peripheral blood mononuclear cells (PBMCs), 3’UTR shortening in activated and proliferating cardiac fibroblasts from injured mouse hearts and finally by building an initial atlas of cell type-specific transcript usage across 12 mouse tissues.
Conclusions We anticipate that Sierra will enable new avenues of transcriptional complexity and regulation to be explored in single-cell transcriptomic experiments. Sierra is available at https://github.com/VCCRI/Sierra.