Three-dimensional topology of the genome: A computational modeling perspective

Abstract

The genome has long been believed to be more or less randomly arranged in the nucleus. However, recent advancements in biochemical, microscopy and computational techniques have revealed a hierarchical genome organization with a vital role for the life and development of an organism. Critically, a disorganization might lead to pathological conditions. To understand nuclear architecture, it is critical to determine which regions of the genome are spatially close to each other or proximal to compartments such as the nuclear lamina, at the nuclear periphery. Using chromosome conformation capture coupled with high-throughput sequencing (Hi-C), genome-wide contact frequencies can be obtained. Further, with chromatin immunoprecipitation sequencing (ChIP-seq), regions interacting with nuclear compartments can be identified. Such information is crucial because spatial positioning affects gene expression regulation. For example, genes residing close to the nuclear periphery tend to be repressed, whereas genes in the nuclear center are generally expressed. In my PhD work, I combined such data to develop a computational pipeline to generate 3D genome structural models that predict the positions of chromatin in the 3D nucleus. These computational models can be used study the position of genomic regions of interest in 3D space. I have also investigated the 4-dimensional aspect of genome organization, that is, how 3D genome conformation changes over time. I used Hi-C and ChIP-seq data from stem cells at a different time points of adipose differentiation. We discovered a new level of genome organization wherein topological chromatin domains dynamically associate in space during differentiation, forming ‘cliques’ enriched in repressed genomic regions. Lastly, I have characterized these cliques in different human cell types to reveal characteristic features of their boundaries and how these relate to epigenetic modifications.