New software tools for differential analysis of single-cell genomics perturbation experiments

Single-cell genomics technology has advanced at a blistering pace. The throughput of single-cell transcriptome sequencing has increased by four orders of magnitude in the past five years alone, enabling our group and others to catalog all of the cell types in a whole embryo within a single experiment. In parallel, assays for diverse aspects of the epigenome, including chromatin accessibility, DNA methylation, and histone modifications have been adapted to work in single cells and at scale. Furthermore, multiplexing techniques have raised the prospect of using single-cell genomics not only to catalog cell types, but to comprehensively study the effects of myriad perturbations of embryonic development, or to characterize the evolution of disease pathogenesis at whole-animal scale and molecular resolution. In principle, single-cell genomics could serve as an extraordinarily high-content means of phenotyping, but the volume and richness of datasets produced by such experiments poses new, daunting computational and statistical challenges. A lack of software tools for comparing specimens profiled as part of single-cell RNA-seq or ATAC-seq experiments constitutes a critical gap in the field. This proposal aims to fill that gap with software tools that will allow users to characterize how disease progression, genetic or chemical perturbations, or environmental effects alter the proportions and molecular states of cells in complex tissues or whole embryos. In order to establish the accuracy of our tools and the physiological relevance of their predictions, we will extensively validate their output through analysis of existing and newly generated single-cell sequencing data using the very tractable zebrafish embryonic development system. In our first Aim, we will develop software for detecting shifts within cell populations across healthy and pathological molecular states. In our second Aim, we will develop software for identifying genes that mediate or regulate cell-state transitions during development or disease pathogenesis. In our third Aim, we will develop methods for defining how chromatin states at regulatory DNA control transcriptional states. Upon completing these aims, we will have delivered new, widely applicable software for analyzing single-cell genomics experiments. We will also have produced new datasets that will serve both as a reference map for vertebrate embryogenesis and a platform for further development of tools for genetic analysis by our group and others. Our experiments will also yield new insights as to how vertebrate genomes encode developmental programs.