INTEGRATIVE ANALYSIS OF SINGLE CELL MULTI-OMICS DATA

Abstract

Development of single cell sequencing technologies have made it possible to quantify the genetic materials from an individual cell, providing great potential for understanding both the heterogeneity of biological systems and the general principles of biological processes. Various assays have been established to measure the transcriptome, chromatin accessibility, DNA methylation and certain histone modifications at the resolution of single cells. How to combine the multi-omic collection of single cell assays to understand the molecular basis of biological process is an active research area. In this dissertation we present our work on interpreting the regulation of gene expression in a data-driven way from these multi-omic single cell datasets and developing new bioinformatics tools for this purpose. We analyzed the single nucleus transcriptome, chromatin accessibility and methylation profiles from the pituitary tissue in an integrative manner. We extracted transcription factors and cis-regulatory regions from the datasets. We then further integrated these analyses to show a multilayer model of gene expression regulation involving both the presence of transcription factors and the pattern of chromatin structure. To further understand the role of chromatin accessibility in gene expression regulation, we analyzed a same-cell multi-omics dataset where both the transcriptome and the chromatin accessibility were measured in the same single cells. We discovered a diversity of mechanisms contributing to the differential expression of genes, and different genes were dominated by different mechanisms. Finally, we built a general computational tool to extract regulatory mechanisms including transcription factors and regulatory domains from the same-cell multi-omics assay of single cells. We showed that this tool gave validated predictions and provided insights into the cell type-specific expression of genes. Together these effort provided a general framework for elucidating the mechanisms underlying biological processes from multi-omic single cell datasets.