19.1: Introduction

The human body contains approximately 210 different cell types, but each cell type shares the same genomic sequence. In spite of having the same genetic code, cells not only develop into distinct types from this same sequence, but also maintain the same cell type over time and across divisions. This information about the cell type and the state of the cell is called epigenomic information. The epigenome (“epi” means above in Greek, so epigenome means above genome) is the set of chemical modifications or marks that influence gene expression and are transferred across cell divisions and, in some limited cases, across generations of organisms.

As shown in Figure 19.1, epigenomic information in a cell is encoded in diverse ways. For example, methylation of DNA (e.g. at CpG dinucleotides) can alter gene expression. Similarly, positioning of nu- cleosomes (unit of packing of DNA) determines which parts of DNA are accessible for transcription factors to bind to and other enzymes. Almost two decades of work have revealed hundreds of post translational modifications of histone tails. Since an extremely large number of histone modification states are possible for any given histone tail, the ”histone code hypothesis” has been proposed. This hypothesis states that particular combinations of histone modifications encode information. Although a controversial hypothesis, it has guided the epigenetics field. The core of epigenetics is understanding how chemical modifications to chromatin (be they DNA methylation, histone modifications or chromatin architecture) are established and how the cell ”interprets” this information to establish and maintain gene expression states.

In this chapter we will explore the experimental and computational techniques used to uncover chromatin states within a cell type. We will learn how chromatin immunoprecipitation can be used to infer the regions of the genome bound by a protein or interest, and a common algorithm (the Burrows-Wheeler) transform can be used to rapidly map large numbers of short sequencing reads to a reference genome. From this we then abstract a level and use a hidden Markov model (HMM) to segment the genome into regions which share similar chromatin states. We will close by showing how these comprehensive maps of chromatin states can be compared across cell types and can be used to provide information on how cell states are established and maintained and the impact of genetic variation on gene expression.