Detection of differentially abundant cell subpopulations in scRNA-seq data

Significance Comparative analysis of samples from two biological states, such as two stages of embryonic development, is a pressing problem in single-cell RNA sequencing (scRNA-seq). A key challenge is to detect cell subpopulations whose abundance differs between the two states. To that end, we develop DA-seq, a multiscale strategy to compare two cellular distributions. In contrast to existing unsupervised clustering-based analysis, DA-seq can delineate cell subpopulations with the most significant discrepancy between two states and potentially reveal important changes in cellular processes that are undetectable using conventional methods.