College of Science
56 Interpreting the Consequences of Single-Cell CRISPR Perturbations Through Label Transfer
Tejashree Prakash (University of Utah); Clayton M. Carey (University of Utah); Shengzhou Wang (University of Utah); and James A. Gagnon (School of Biological Sciences, University of Utah)
Faculty Mentor: James A. Gagnon (School of Biological Sciences, University of Utah)
Single-cell RNA sequencing (scRNA-seq) permits the measurement of gene expression at single cell resolution, enabling a new understanding of the cell and molecular responses to perturbations in developing animals. However, a challenge for using scRNA-seq data exists in defining shared cell types across samples. My objective is to develop an agnostic computational framework to transfer cell type labels onto a perturbed sample and analyze how gene expression and cell type abundances are affected. To test this label transfer method, I focused on the gene regulatory network that establishes mesoderm cell types in the zebrafish embryo. The genes of interest in this network, noto, tbxta, and tbx16, are wellknown mesoderm regulators, but have not been studied at single-cell resolution. Using CRISPR, I perturbed each transcription factor and generated scRNA-seq libraries for each mutant at 24hpf. Each condition’s gene expression profiles were processed and integrated through Seurat’s clustering workflow to aggregate the cell populations present in all conditions. Marker genes for each cluster of the integrated object, with all perturbed conditions, act as the input for the annotation framework to identify the most likely cell type for each cluster. By employing this framework, I identified the different cell types present across each perturbation condition in the developing embryos. Preliminary analysis of the labeled singlecell datasets suggests changes in cell abundance of cell types previously known, such as changes in notochord and muscle. Further analysis and experiments are required to validate the cell type abundance changes seen in these datasets. The annotation framework has allowed me to begin studying mesoderm patterning interactions between noto, tbxta, and tbx16 at a molecular level. Due to the agnostic nature of this computational framework, this method is broadly useful for studying cell and gene expression in other single-cell perturbations.