Computational approaches to integrate epigenomics and transcriptomics in differentiation and in disease

Abstract

The development of next generation sequencing (NGS) methods, in conjunction with bioinformatical methods, has opened unprecedented opportunities for studying epigenetics. Indeed, due to the recent explosion of epigenome datasets produced with the help of NGS, computational methods are being increasingly applied in all areas of epigenetics research. The present dissertation focuses on the goal of integrating epigenomic with transcriptome datasets with the help of computational methods. I will illustrate the utility of this approach in two different contexts. First, I used epigenomics data obtained from differentiating Neural Stem Cells (NSCs) in order to identify cis-regulators (enhancers) of gene transcription. Second, I built a novel database, called EPITRANS, which provides a framework to identify new classes of disease-related epigenetic regulators. Identification of enhancers in differentiating NSCs (Chapter 2). Apart from transcription factor (TF) availability, functionality of the transcriptional machinery depends on epigenetic modifications, which influence chromatin conformation, and on cis-regulatory elements such as enhancers, which provide TF binding sites. Enhancers are often difficult to identify due to their distances from their regulated genes. In the present dissertation, NSCs from mouse embryos (embryonic day 14.5 (E14.5)) were differentiated for 1 (DF-1) or 7 (DF-7) days. These NSCs normally differentiate into all cell lineages in the central nervous system. The gene expression patterns in the DF-1 and DF-7 cells were analyzed by RNA-Seq, while putative cis-regulatory elements (enhancers) were identified by histone modification ChIP-Seq. We found 2210 peaks in p300 binding sites based on the histone modification ChIP-seq results, and identified a total of 420 putative active enhancers in the overlap of H3K4me1 and H3K27ac. 491 differentially expressed genes (DEGs) in RNA-seq data were selected and based on gene ontology classification, the DEGs identified were related to developmental processes and cell adhesion. Construction of EPITRANS (Chapter 3). The biological database EPITRANS was created using publicly available data on human epigenetic marks and human disease. This database reveals relationships between epigenetic modifications and gene expression data and displays them visually for each human cell type. In this way EPITRANS facilitates the integration of different resources and provides a new framework to identify novel classes of epigenetic regulators related with disease. Although EPITRANS so far integrates information only from selected open-access biological and biomedical resources, it is flexible enough to integrate additional epigenomic and publicly available biological data. A future goal is to expand the text mining and integration of experimental data with biological pathways, disease networks and structural biology using the NGS technology.