The SpliZ generalizes “Percent Spliced In” to reveal regulated splicing at single-cell resolution

Abstract

AbstractTo date, detecting robust single-cell-regulated splicing is viewed as out of reach from droplet based technologies such as 10x Chromium. This prevents the discovery of single-cell-regulated splicing in rare cell types or those that are difficult or impossible to sequence deeply. Here, we introduce a novel, robust, and computationally efficient set of statistics, the Splicing Z Score (SpliZ) and SpliZVD, to detect regulated splicing in single cell RNA-seq including 10x Chromium. The SpliZ(VD) provides annotation-free detection of differentially regulated, complex alternative splicing events. The SpliZ generalizes and increases statistical power compared to the Percent Spliced In (PSI) and mathematically reduces to PSI for simple exon-skipping. We applied the SpliZ to primary human lung cells to discover hundreds of genes with new regulated cell-type-specific splicing. The SpliZ has wide application to enable biological discovery of genes predicted to have functionally significant splicing programs including those regulated in development.