Ernst J, Kellis M.ChromHMM: automating chromatin-state discovery and characterization. Nat Methods 9:215-216

1] Computer Science and Artificial Intelligence Laboratory, Massachusetts Institute of Technology, Cambridge, Massachusetts, USA. [2] Broad Institute of Massachusetts Institute of Technology and Harvard, Cambridge, Massachusetts, USA. [3] Department of Biological Chemistry, University of California Los Angeles, Los Angeles, California, USA.
Nature Methods (Impact Factor: 32.07). 02/2012; 9(3):215-6. DOI: 10.1038/nmeth.1906
Source: PubMed


To the Editor:
Chromatin-state annotation using combinations of chromatin modification patterns has emerged as a powerful approach for discovering regulatory regions and their cell type–specific activity patterns and for interpreting disease-association studies1, 2, 3, 4, 5. However, the computational challenge of learning chromatin-state models from large numbers of chromatin modification datasets in multiple cell types still requires extensive bioinformatics expertise. To address this challenge, we developed ChromHMM, an automated computational system for learning chromatin states, characterizing their biological functions and correlations with large-scale functional datasets and visualizing the resulting genome-wide maps of chromatin-state annotations.

Download full-text


Available from: Manolis Kellis, Oct 05, 2014
81 Reads
  • Source
    • "Combinations of epigenetic marks can be grouped into chromatin states, which describe the most probable combinations of shared chromatin peaks that define local epigenetic environments. We used ChromHMM (Ernst and Kellis 2012) (Table S3, Figure S2 and Methods) to find an eleven-state model that represents the ten most frequent combinations of modifications in our dataset, together with a zero state containing no modified histone peaks. In Figure 3b we have ordered and color-coded these states by their decreasing involvements with active chromatin histone marks. "
    [Show abstract] [Hide abstract]
    ABSTRACT: Combinations of histones carrying different covalent modifications are a major component of epigenetic variation. We have mapped nine modified histones in the barley seedling epigenome by ChIP-seq. The chromosomal distributions of the modifications group them into four different classes and members of a given class also tend to coincide at the local DNA level, suggesting that global distribution patterns reflect local epigenetic environments. We used this peak sharing to define ten chromatin states representing local epigenetic environments in the barley genome. Five states map mainly to genes and five to intergenic regions. Two genic states involving H3K36me3 are preferentially associated with constitutive gene expression, while an H3K27me3-containing genic state is associated with differentially-expressed genes. The ten states display striking distribution patterns that divide barley chromosomes into three distinct global environments. First, telomere-proximal regions contain high densities of H3K27me3 covering both genes and intergenic DNA, together with very low levels of the repressive H3K27me1 modification. Flanking these are gene-rich interior regions that are rich in active chromatin states, greatly decreased levels of H3K27me3 and increasing amounts of H3K27me1 and H3K9me2. Lastly, H3K27me3-depleted pericentromeric regions contain gene islands with active chromatin states separated by extensive retrotransposon-rich regions that are associated with abundant H3K27me1 and H3K9me2 modifications. We propose an epigenomic framework for barley whereby intergenic H3K27me3 specifies facultative heterochromatin in the telomere-proximal regions and H3K27me1 is diagnostic for constitutive heterochromatin elsewhere in the barley genome. This article is protected by copyright. All rights reserved. This article is protected by copyright. All rights reserved.
    The Plant Journal 08/2015; 84(1). DOI:10.1111/tpj.12963 · 5.97 Impact Factor
  • Source
    • "We observed two broad classes of prospective myeloid enhancers in pre- B cells as follows: (1) pre-existing enhancers that were decorated with H3K4Me1, H3K27Ac, and P300 and depleted for H3K27Me3 (Figure 2A); and (2) de novo enhancers that lacked any of the active enhancer marks but were instead often decorated with H3K27Me3 (Figure 2A). Similar results were obtained by performing ChromHMM analysis (Figure S2C; Ernst and Kellis, 2012) as an independent analytical approach demonstrating that pre-existing enhancers are enriched for activation marks, whereas de novo enhancers are depleted for activation marks and enriched for H3K27Me3 (Figure S2D). Approximately twothirds of the pre-existing enhancers were bound by PU.1 and exhibited high levels of activation marks compared to sites not bound by PU.1 (Figure S2B). "
    [Show abstract] [Hide abstract]
    ABSTRACT: Transcription-factor-induced somatic cell conversions are highly relevant for both basic and clinical research yet their mechanism is not fully understood and it is unclear whether they reflect normal differentiation processes. Here we show that during pre-B-cell-to-macrophage transdifferentiation, C/EBPα binds to two types of myeloid enhancers in B cells: pre-existing enhancers that are bound by PU.1, providing a platform for incoming C/EBPα; and de novo enhancers that are targeted by C/EBPα, acting as a pioneer factor for subsequent binding by PU.1. The order of factor binding dictates the upregulation kinetics of nearby genes. Pre-existing enhancers are broadly active throughout the hematopoietic lineage tree, including B cells. In contrast, de novo enhancers are silent in most cell types except in myeloid cells where they become activated by C/EBP factors. Our data suggest that C/EBPα recapitulates physiological developmental processes by short-circuiting two macrophage enhancer pathways in pre-B cells. Copyright © 2015 The Authors. Published by Elsevier Inc. All rights reserved.
    Stem Cell Reports 07/2015; 5(2). DOI:10.1016/j.stemcr.2015.06.007 · 5.37 Impact Factor
  • Source
    • "Less than one third of Ascl1 binding occurs inside genes (30%) or their promoter regions (7%) (Figures 2A and 2B), suggesting that Ascl1 binds predominantly to distal enhancer regions. We next used a Hidden-Markov model to characterize the chromatin states at Ascl1-bound regions, using genome-wide profiles of histone modifications generated from proliferating and differentiating NS cells (before or 24 hr after addition of tamoxifen, respectively ) (Ernst and Kellis, 2012). Ascl1 BEs fall mostly within regions of chromatin highly co-enriched for H3K4me1 and H3K27ac, characteristic of active enhancers (same result for p < 10 À10 ; data not shown). "
    [Show abstract] [Hide abstract]
    ABSTRACT: The proneural transcription factor Ascl1 coordinates gene expression in both proliferating and differentiating progenitors along the neuronal lineage. Here, we used a cellular model of neurogenesis to investigate how Ascl1 interacts with the chromatin landscape to regulate gene expression when promoting neuronal differentiation. We find that Ascl1 binding occurs mostly at distal enhancers and is associated with activation of gene transcription. Surprisingly, the accessibility of Ascl1 to its binding sites in neural stem/progenitor cells remains largely unchanged throughout their differentiation, as Ascl1 targets regions of both readily accessible and closed chromatin in proliferating cells. Moreover, binding of Ascl1 often precedes an increase in chromatin accessibility and the appearance of new regions of open chromatin, associated with de novo gene expression during differentiation. Our results reveal a function of Ascl1 in promoting chromatin accessibility during neurogenesis, linking the chromatin landscape at Ascl1 target regions with the temporal progression of its transcriptional program. Copyright © 2015 The Authors. Published by Elsevier Inc. All rights reserved.
    Cell Reports 03/2015; 10(9):1-13. DOI:10.1016/j.celrep.2015.02.025 · 8.36 Impact Factor
Show more

Similar Publications