Abstract

A plethora of epigenetic modifications have been described in the human genome and shown to play diverse roles in gene regulation, cellular differentiation, and the onset of disease. While some modifications have been linked with activity levels of different functional elements, their combinatorial patterns remain unresolved, and their potential for systematic de novo genome annotation remains untapped. In this paper, we systematically discover and characterize recurrent spatially-coherent and biologically-meaningful chromatin mark combinations, or chromatin states, in human T-cells. We describe 51 distinct chromatin states, including promoter-associated, transcription-associated, active intergenic, large-scale repressed and repeat-associated states. Each chromatin state shows specific functional, experimental, conservation, annotation, and sequence-motif enrichments, revealing their distinct candidate biological roles. Overall, our work provides a complementary functional annotation of the human genome revealing the genome-wide locations of diverse classes of epigenetic functions, including previously-unsuspected chromatin states enriched in transcription end sites, distinct repeat families, and disease-SNP-associated states. While the primary DNA sequence of the human genome is ultimately responsible for the