In mammals the genome is shaped by epigenetic regulation to manifest numerous cellular identities. The term epigenetics has been used to refer to changes in gene expression, which are heritable through multiple cell division cycles that are not due to variations in primary DNA sequence. Stable suppression of differentiation genes is required to sustain the undifferentiated state in cells ranging from embryonic stem cells to somatic stem cell progenitors that constantly replenish self-renewing tissues. However, the epigenetic mechanisms behind the maintenance of cellular dedifferentiation are not yet fully understood. Major effectors of epigenetic control include regulators of DNA methylation and histone modification as well as ATP-dependent chromatin remodeling enzymes. These interact with other regulators, such as DNA sequence-specific transcription factors and noncoding RNAs to landscape the genome during development, differentiation and cancer. DNA methylation is a classic and powerful example of the epigenetic inheritance of cellular identity that is widely used in eukaryotes. DNA methylation confers distinct epigenetic states via several mechanisms. Here we discuss fundamental mechanisms of DNA methylation and their interplay with several regulatory pathways that define cellular physiology and differentiation.

Next-generation sequencing technologies provide a powerful tool for studying genome evolution during progression of advanced diseases such as cancer. Although many recent studies have employed new sequencing technologies to detect mutations across multiple, genetically related tumors, current methods do not exploit available phylogenetic information to improve the accuracy of their variant calls. Here, we present a novel algorithm that uses somatic single-nucleotide variations (SNVs) in multiple, related tissue samples as lineage markers for phylogenetic tree reconstruction. Our method then leverages the inferred phylogeny to improve the accuracy of SNV discovery. Experimental analyses demonstrate that our method achieves up to 32% improvement for somatic SNV calling of multiple, related samples over the accuracy of GATK's Unified Genotyper, the state-of-theart multisample SNV caller.

Next-generation sequencing technologies provide a powerful tool for studying genome evolution during progression of advanced diseases such as cancer. Although many recent studies have employed new sequencing technologies to detect mutations across multiple, genetically related tumors, current methods do not exploit available phylogenetic information to improve the accuracy of their variant calls. Here, we present a novel algorithm that uses somatic single-nucleotide variations (SNVs) in multiple, related tissue samples as lineage markers for phylogenetic tree reconstruction. Our method then leverages the inferred phylogeny to improve the accuracy of SNV discovery. Experimental analyses demonstrate that our method achieves up to 32% improvement for somatic SNV calling of multiple, related samples over the accuracy of GATK's Unified Genotyper, the state-of-theart multisample SNV caller.