Single-cell allele-specific expression analysis reveals dynamic and cell-type-specific regulatory effects

Allele-specific expression, which measures the expression of two alleles of a gene in a diploid individual, is a powerful signal to study cis-regulatory effects. Comparing ASE across conditions, or differential ASE, can reveal context-specific gene regulation. Recently, single-cell RNA sequencing (scRNA-seq) has allowed the measurement of ASE at the resolution of individual cells, but there is a lack of statistical methods to analyze such data. We develop DAESC, a statistical method for differential ASE analysis across any condition of interest using scRNA-seq data from multiple individuals. DAESC includes a baseline model based on beta-binomial regression with random effects accounting for multiple cells from the same individual (DAESC-BB), and an extended mixture model that incorporates implicit haplotype phasing (DAESC-Mix). We demonstrate through simulations that DAESC accurately captures differential ASE effects in a wide range of scenarios. Application to scRNA-seq data from 105 induced pluripotent stem cell lines identifies 657 genes that are dynamically regulated during endoderm differentiation. A second application identifies several genes that are differentially regulated in pancreatic endocrine cells between type 2 diabetes patients and controls. In conclusion, DAESC is a powerful method for single-cell differential ASE analysis and can facilitate the discovery of context-specific regulatory effects.