| The study of chromosome, and genome, organization is a both an ongoing challenge, and one with a long history. Following the advent of high-throughput sequencing and genomic technologies, much research has focused on the one-dimensional, or sequence-level, organization of genomes, with many successes. Nonetheless, genomes are physically organized as chromosomes in the three-dimensional confines of the cell nucleus, with implications for processes including gene regulation, DNA replication, and cell division. Recently, chromosome conformation capture (3C) based methods have enabled new high-resolution and genome-wide views (Hi-C) of chromosome organization in three-dimensions. 3C methods convert direct spatial contacts between pairs of genomic loci into molecular products that can be assayed using high-throughput sequencing. The new views of chromosomal organization enabled by 3C techniques have been the principal motivation for my graduate research. In particular, 3C technologies now pose multiple important computational and theoretical challenges, including how to: (1) process and filter large quantities of experimental data; (2) develop computational models of chromosomes that agree with and help the understanding of experimental data; and (3) integrate views from 3C technologies with other genomic datasets, including complementary characterizations of the chromatin fiber. This thesis presents a series of projects addressing these challenges in chronological order of their publication. The first project relates to the integration of views from Hi-C with other genomic datasets to understand the functional implications of chromosome organization. This project examined the connection between Hi-C chromosome contact maps and the distribution and positions of somatic copy number alterations observed across a variety of cancers. Since the observed alterations are the consequence of both mutational processes and evolutionary pressures on the cancers, we used a population genetics framework to consider how a mutational process governed by polymer physics might manifest in the patterns of alterations observed in cancer genomes. The second project relates to the processing and filtering of Hi-C data. This project investigated how to correct for various biases that could be introduced at different stages of the experimental protocol, and then how to decompose the resulting contact maps into dominant features of chromosomal organization. We found that we could dramatically compress the complexity of chromosome interaction patterns, and that these compressed patterns are surprisingly conserved between humans and mice. These methods have been used by our lab and others to investigate 3C data across a broad range of organisms. The third project involved the analysis of Hi-C data through the cell cycle, and the development of polymer models of chromosome organization in metaphase, when cells are prepared for division. Before cell division, chromosomes undergo extensive compaction; after division, they decondense and resume their cell-type-specific gene expression in interphase. We found that while interphase chromosome organization reflects cell-type-specific programs of gene expression, all traces of this organization are wiped clear in metaphase chromosomes. Our models of metaphase chromosomes allowed us to discriminate between two classic biological hypotheses of metaphase chromosome organization. We found that metaphase chromosomes are inconsistent with classic hierarchical models of folding, yet can be described by a two-stage process of compaction. The fourth project used polymer models to understand how local interactions, or loops, between genomic elements might in turn alter local chromosome organization. This has implications for gene regulation, as the classic model of eukaryotic gene expression requires direct spatial contact between a distal enhancer and a proximal promoter. We found that a chromatin loop can either suppress or facilitate enhancer-promoter interactions, depending on the location of the loop relative to the enhancer-promoter pair, and that looping interactions that do not directly involve an enhancer-promoter pair can nevertheless significantly modulate their interactions. |
