UC Berkeley

Alternative splicing is a post-transcriptional gene regulatory process that increases transcriptomic diversity through the differential selection of exons. Over the past decade, single cell RNA sequencing (scRNA-seq) experiments have allowed the study of the complex heterogeneity of tissues, and the gene expression regulation mechanisms that determine the identity of individual cells. However the study of alternative splicing has largely lagged behind in single-cell approaches. This is in part due to a number of technical limitations that make this study particularly difficult. In this dissertation, I first demonstrate that the technical distortion of splicing observations in single cells produces the appearance of bimodality in single cell splicing. I discuss how this limitation has affected our understanding of single cell splicing, and what computational approaches can be used to extract meaningful biological information from this exceptionally noisy data. I then present Psix: a computational tool that implements an updated model for single cell splicing observations that takes into account this distortion. Using Psix, I identify alternative splicing that changes across a phenotypic landscape, as well as groups of potentially co-spliced exons. Finally, I explore potential regulatory roles for these poison exons in the production of Serine/Arginine-rich (SR) splicing factor proteins.