UC San Diego

The three-dimensional structure of the genome plays a key role in gene regulation. For example, while highly compacted heterochromatin drives gene silencing, open euchromatin facilitates gene activation. Nevertheless, how chromatin folds within these structures and consequently how it controls access to genomic content is poorly understood. Recent advances in high-throughput sequencing have provided valuable tools, such as Hi-C, for the study of chromatin structure. Using Hi-C datasets, I developed a hidden markov based algorithm to identify self-interacting patterns of chromatin structure termed topological domains. These mega-base sized domains are pervasive throughout the genome and are highly conserved among humans and mouse. At a higher resolution, topological domains encompass individual chromatin interactions between regulatory elements and its target gene. Therefore, in order to mechanistically understand gene regulation, it is essential to elucidate the functional relationship among regulatory elements and their target genes. By exploiting the sequence diversity between homologous chromosomes, it is possible to delineate this relationship. However, this requires the knowledge of haplotypes, which has traditionally been difficult to obtain. As the Hi-C protocol preferentially recovers DNA variants on the same chromosome, I invented HaploSeq to reconstruct chromosome-scale haplotypes. HaploSeq can generate haplotypes with ̃99.5% accuracy for >95% of alleles in mouse and 98% accuracy for ̃81% of alleles in humans, thus solving a long-standing problem in genetics. By integrating the knowledge of haplotypes, we queried the relationship between regulatory elements and gene expression in human embryonic stem cells and a panel of differentiated cell-types. Across the 5 cell lineages examined, I identified a total of 24% of genes that showed allelic bias in gene expression. While most of the allelic -genes had a correlating allelic-promoter chromatin state, ̃29% of genes were exceptions suggesting other mechanisms of gene regulation. Accordingly, I then analyzed histone- acetylation marks to identify 1589 allelic enhancers. By predicting chromatin interactions using Hi-C, we observed allelic enhancers to be spatially proximal to allelic genes, suggesting cooperative activity among genome sequence, structure, and function. Taken together, our studies suggest that gene regulation is facilitated and coordinated by genome structure