Identifying genes with cell-type-specific alternative polyadenylation in multi-cluster single-cell transcriptomics data

ABSTRACT Alternative polyadenylation (APA) causes widespread shortening or lengthening of the 3′-untranslated region (3′-UTR) of genes across multiple cell types (dynamic APA). Bioinformatic tools have been developed to identify dynamic APA in single cell RNA-Seq (scRNA-Seq) data, but they suffer from low power and cannot identify APA genes specific to each cell type (cell-type-specific APA) when multiple cell types are analyzed. To address these limitations, we develop a model-based method, scMAPA. scMAPA quantifies 3′-UTR long and short isoforms without posing assumptions on the signal shape of input data, enabling a sensitive identification of APA genes. In human peripheral blood mono cellular data, this enhanced power identifies unique associations of dynamic APA with hematological processes, e.g. progenitor cell development. Further, scMAPA identifies APA genes specific to each cell type while controlling confounders using a sophisticated statistical model. In mouse brain data, scMAPA identifies APA genes specific to each cell type and provides a novel implication of neuron-specific APA genes in the interaction between neurons and blood vessels. Altogether, scMAPA sheds novel insights into the function of cell-type-specific APA dynamics in complex tissues.