Modeling and integration of multi-omics data to study regulatory landscapes governing placenta development

Abstract

The placenta is a transient organ that is crucial during pregnancy. It has multiple functions to ensure optimal fetal growth, including nutrient transport, oxygen exchange and immune protection. The placenta develops in a stage-wise manner and requires precise regulation of gene expression. Abnormalities in placental gene regulation can lead to pregnancy disorders such as preeclampsia, placenta accreta and placental abruption, which can be detrimental to the short and long-term health of both the mother and the fetus. However, the regulatory mechanisms governing placental development, especially with respect to gene regulatory networks, are poorly understood. In this dissertation, we aimed to identify regulatory networks associated with placental development by developing computational methods, and by analyzing and integrating various sequencing data at both the bulk and single-cell level. First, we generated and analyzed transcriptomic data from mouse fetal placenta tissues at embryonic day (e) 7.5, e8.5 and e9.5 to identify groups of genes that regulate placenta-specific developmental processes using cluster analysis, differential expression analysis, and network analysis. Second, we developed a deep learning framework to identify genome-wide chromatin accessibility regions. This framework is applicable for not only for placenta-derived data but also data generated in other tissues. Third, we integrated single-cell transcriptome and single-nucleus chromatin accessibility data generated from the rat uterine-placental interface to identify conserved gene regulatory networks governing rat and human placenta development. The completion of these studies has led to a better understanding of the gene – gene, gene – transcription factor, and transcription factor – cis-regulatory element interactions regulating placental development. Furthermore, the pipelines and tools developed, including the novel deep learning framework for chromatin accessibility analysis, are not limited to rodent and human placenta, but can be used to analyze data generated in any tissue or organism.