Nature 11247

ARTICLE
doi:10.1038/nature11247
An integrated encyclopedia of DNA elements in the human genome

The ENCODE Project Consortium*
The human genome encodes the blueprint of life, but the function of the vast majority of its nearly three billion bases is unknown. The Encyclopedia of DNA Elements (ENCODE) project has systematically mapped regions of transcription, transcription factor association, chromatin structure and histone modification. These data enabled us to assign biochemical functions for 80% of the genome, in particular outside of the well-studied protein-coding regions. Many discovered candidate regulatory elements are physically associated with one another and with expressed genes, providing new insights into the mechanisms of gene regulation. The newly identified elements also show a statistical correspondence to sequence variants linked to human disease, and can thereby guide interpretation of this variation. Overall, the project provides new insights into the organization and regulation of our genes and genome, and is an expansive resource of functional annotations for biomedical research. The human genome sequence provides the 95% of the genome lies within 8 kilobases (kb) of a DNAprotein interaction (as assayed by underlying code for human biology. Despite bound ChIP-seq motifs or DNase I footprints), intensive study, especially in identifying Encyclopedia of DNA Elements and 99% is within 1.7 kb of at least one of the protein-coding genes, our understanding of the nature.com/encode biochemical events measured by ENCODE. genome is far from complete, particularly with regard to non-coding RNAs, alternatively spliced transcripts and reg- . Primate-specific elements as well as elements without detectable ulatory sequences. Systematic analyses of transcripts and regulatory mammalian constraint show, in aggregate, evidence of negative selecinformation are essential for the identification of genes and regulatory tion; thus, some of them are expected to be functional. regions, and are an important resource for the study of human biology . Classifying the genome into seven chromatin states indicates an initial and disease. Such analyses can also provide comprehensive views of the set of 399,124 regions with enhancer-like features and 70,292 regions organization and variability of genes and regulatory information across with promoter-like features, as well as hundreds of thousands of quiescent regions. High-resolution analyses further subdivide the genome cellular contexts, species and individuals. The Encyclopedia of DNA Elements (ENCODE) project aims to into thousands of narrow states with distinct functional properties. delineate all functional elements encoded in the human genome13. . It is possible to correlate quantitatively RNA sequence production Operationally, we define a functional element as a discrete genome and processing with both chromatin marks and transcription factor segment that encodes a defined product (for example, protein or binding at promoters, indicating that promoter functionality can non-coding RNA) or displays a reproducible biochemical signature explain most of the variation in RNA expression. (for example, protein binding, or a specific chromatin structure). . Many non-coding variants in individual genome sequences lie in Comparative genomic studies suggest that 38% of bases are under ENCODE-annotated functional regions; this number is at least as purifying (negative) selection48 and therefore may be functional, large as those that lie in protein-coding genes. although other analyses have suggested much higher estimates911. . Single nucleotide polymorphisms (SNPs) associated with disease by In a pilot phase covering 1% of the genome, the ENCODE project GWAS are enriched within non-coding functional elements, with a annotated 60% of mammalian evolutionarily constrained bases, but majority residing in or near ENCODE-defined regions that are outalso identified many additional putative functional elements without side of protein-coding genes. In many cases, the disease phenotypes evidence of constraint2. The advent of more powerful DNA sequencing can be associated with a specific cell type or transcription factor. technologies now enables whole-genome and more precise analyses ENCODE data production and initial analyses with a broad repertoire of functional assays. Here we describe the production and initial analysis of 1,640 data Since 2007, ENCODE has developed methods and performed a large sets designed to annotate functional elements in the entire human number of sequence-based studies to map functional elements across genome. We integrate results from diverse experiments within cell types, the human genome3. The elements mapped (and approaches used) related experiments involving 147 different cell types, and all ENCODE include RNA transcribed regions (RNA-seq, CAGE, RNA-PET and data with other resources, such as candidate regions from genome-wide manual annotation), protein-coding regions (mass spectrometry), association studies (GWAS) and evolutionarily constrained regions. transcription-factor-binding sites (ChIP-seq and DNase-seq), Together, these efforts reveal important features about the organization chromatin structure (DNase-seq, FAIRE-seq, histone ChIP-seq and and function of the human genome, summarized below. MNase-seq), and DNA methylation sites (RRBS assay) (Box 1 lists . The vast majority (80.4%) of the human genome participates in at methods and abbreviations; Supplementary Table 1, section P, details least one biochemical RNA- and/or chromatin-associated event in at production statistics)3. To compare and integrate results across the least one cell type. Much of the genome lies close to a regulatory event: different laboratories, data production efforts focused on two selected
ENCODE
Lists of participants and their affiliations appear at the end of the paper. 6 S E P T E M B E R 2 0 1 2 | VO L 4 8 9 | N AT U R E | 5 7
2012 Macmillan Publishers Limited. All rights reserved
RESEARCH ARTICLE
BOX 1
dataStandards.html; A. Kundaje, personal communication). Uniform data-processing methods were developed for each assay (see Supplementary Information; A. Kundaje, personal communication), and most assay results can be represented both as signal information (a per-base estimate across the genome) and as discrete elements (regions computationally identified as enriched for signal). Extensive processing pipelines were developed to generate each representation (M. M. Hoffman et al., manuscript in preparation and A. Kundaje, personal communication). In addition, we developed the irreproducible discovery rate (IDR)13 measure to provide a robust and conservative estimate of the threshold where two ranked lists of results from biological replicates no longer agree (that is, are irreproducible), and we applied this to defining sets of discrete elements. We identified, and excluded from most analyses, regions yielding untrustworthy signals likely to be artefactual (for example, multicopy regions). Together, these regions comprise 0.39% of the genome (see Supplementary Information). The poster accompanying this issue represents different ENCODE-identified elements and their genome coverage.
Transcribed and protein-coding regions We used manual and automated annotation to produce a comprehensive catalogue of human protein-coding and non-coding RNAs as well as pseudogenes, referred to as the GENCODE reference gene set14,15 (Supplementary Table 1, section U). This includes 20,687 protein-coding genes (GENCODE annotation, v7) with, on average, 6.3 alternatively spliced transcripts (3.9 different protein-coding transcripts) per locus. In total, GENCODE-annotated exons of proteincoding genes cover 2.94% of the genome or 1.22% for protein-coding exons. Protein-coding genes span 33.45% from the outermost start to stop codons, or 39.54% from promoter to poly(A) site. Analysis of mass spectrometry data from K562 and GM12878 cell lines yielded 57 confidently identified unique peptide sequences in intergenic regions relative to GENCODE annotation. Taken together with evidence of pervasive genome transcription16, these data indicate that additional protein-coding genes remain to be found. In addition, we annotated 8,801 automatically derived small RNAs and 9,640 manually curated long non-coding RNA (lncRNA) loci17. Comparing lncRNAs to other ENCODE data indicates that lncRNAs are generated through a pathway similar to that for protein-coding genes17. The GENCODE project also annotated 11,224 pseudogenes, of which 863 were transcribed and associated with active chromatin18. RNA We sequenced RNA16 from different cell lines and multiple subcellular fractions to develop an extensive RNA expression catalogue. Using a conservative threshold to identify regions of RNA activity, 62% of genomic bases are reproducibly represented in sequenced long (.200 nucleotides) RNA molecules or GENCODE exons. Of these bases, only 5.5% are explained by GENCODE exons. Most transcribed bases are within or overlapping annotated gene boundaries (that is, intronic), and only 31% of bases in sequenced transcripts were intergenic16. We used CAGE-seq (59 cap-targeted RNA isolation and sequencing) to identify 62,403 transcription start sites (TSSs) at high confidence (IDR of 0.01) in tier 1 and 2 cell types. Of these, 27,362 (44%) are within 100 base pairs (bp) of the 59 end of a GENCODE-annotated transcript or previously reported full-length messenger RNA. The remaining regions predominantly lie across exons and 39 untranslated regions (UTRs), and some exhibit cell-type-restricted expression; these may represent the start sites of novel, cell-type-specific transcripts. Finally, we saw a significant proportion of coding and non-coding transcripts processed into steady-state stable RNAs shorter than 200 nucleotides. These precursors include transfer RNA, microRNA, small nuclear RNA and small nucleolar RNA (tRNA, miRNA, snRNA and snoRNA, respectively) and the 59 termini of these processed products align with the capped 59 end tags16.
ENCODE abbreviations
RNA-seq. Isolation of RNA sequences, often with different purification techniques to isolate different fractions of RNA followed by highthroughput sequencing. CAGE. Capture of the methylated cap at the 59 end of RNA, followed by high-throughput sequencing of a small tag adjacent to the 59 methylated caps. 59 methylated caps are formed at the initiation of transcription, although other mechanisms also methylate 59 ends of RNA. RNA-PET. Simultaneous capture of RNAs with both a 59 methyl cap and a poly(A) tail, which is indicative of a full-length RNA. This is then followed by sequencing a short tag from each end by high-throughput sequencing. ChIP-seq. Chromatin immunoprecipitation followed by sequencing. Specific regions of crosslinked chromatin, which is genomic DNA in complex with its bound proteins, are selected by using an antibody to a specific epitope. The enriched sample is then subjected to highthroughput sequencing to determine the regions in the genome most often bound by the protein to which the antibody was directed. Most often used are antibodies to any chromatin-associated epitope, including transcription factors, chromatin binding proteins and specific chemical modifications on histone proteins. DNase-seq. Adaption of established regulatory sequence assay to modern techniques. The DNase I enzyme will preferentially cut live chromatin preparations at sites where nearby there are specific (nonhistone) proteins. The resulting cut points are then sequenced using high-throughput sequencing to determine those sites hypersensitive to DNase I, corresponding to open chromatin. FAIRE-seq. Formaldehyde assisted isolation of regulatory elements. FAIRE isolates nucleosome-depleted genomic regions by exploiting the difference in crosslinking efficiency between nucleosomes (high) and sequence-specific regulatory factors (low). FAIRE consists of crosslinking, phenol extraction, and sequencing the DNA fragments in the aqueous phase. RRBS. Reduced representation bisulphite sequencing. Bisulphite treatment of DNA sequence converts unmethylated cytosines to uracil. To focus the assay and save costs, specific restriction enzymes that cut around CpG dinucleotides can reduce the genome to a portion specifically enriched in CpGs. This enriched sample is then sequenced to determine the methylation status of individual cytosines quantitatively. Tier 1. Tier 1 cell types were the highest-priority set and comprised three widely studied cell lines: K562 erythroleukaemia cells; GM12878, a B-lymphoblastoid cell line that is also part of the 1000 Genomes project (http://1000genomes.org)55; and the H1 embryonic stem cell (H1 hESC) line. Tier 2. The second-priority set of cell types in the ENCODE project which included HeLa-S3 cervical carcinoma cells, HepG2 hepatoblastoma cells and primary (non-transformed) human umbilical vein endothelial cells (HUVECs). Tier 3. Any other ENCODE cell types not in tier 1 or tier 2.
sets of cell lines, designated tier 1 and tier 2 (Box 1). To capture a broader spectrum of biological diversity, selected assays were also executed on a third tier comprising more than 100 cell types including primary cells. All data and protocol descriptions are available at http://www.encodeproject.org/, and a Users Guide including details of cell-type choice and limitations was published recently3.
Integration methodology For consistency, data were generated and processed using standardized guidelines, and for some assays, new quality-control measures were designed (see refs 3, 12 and http://encodeproject.org/ENCODE/
5 8 | N AT U R E | VO L 4 8 9 | 6 S E P T E M B E R 2 0 1 2
ARTICLE RESEARCH
Table 1 | Summary of transcription factor classes analysed in ENCODE
Acronym Description Factors analysed
ChromRem DNARep HISase Other Pol2 Pol3 TFNS TFSS
ATP-dependent chromatin complexes DNA repair Histone acetylation, deacetylation or methylation complexes Cyclin kinase associated with transcription Pol II subunit Pol III-associated General Pol II-associated factor, not site-specific Pol II transcription factor with sequence-specific DNA binding
5 3 8 1 1 (2 forms) 6 8 87
Protein bound regions To identify regulatory regions directly, we mapped the binding locations of 119 different DNA-binding proteins and a number of RNA polymerase components in 72 cell types using ChIP-seq (Table 1, Supplementary Table 1, section N, and ref. 19); 87 (73%) were sequence-specific transcription factors. Overall, 636,336 binding regions covering 231 megabases (Mb; 8.1%) of the genome are enriched for regions bound by DNA-binding proteins across all cell types. We assessed each protein-binding site for enrichment of known DNA-binding motifs and the presence of novel motifs. Overall, 86% of the DNA segments occupied by sequence-specific transcription factors contained a strong DNA-binding motif, and in most (55%) cases the known motif was most enriched (P. Kheradpour and M. Kellis, manuscript in preparation). Protein-binding regions lacking high or moderate affinity cognate recognition sites have 21% lower median scores by rank than regions with recognition sequences (Wilcoxon rank sum P value ,10216). Eighty-two per cent of the low-signal regions have high-affinity recognition sequences for other factors. In addition, when ChIP-seq peaks are ranked by their concordance with their known recognition sequence, the median DNase I accessibility is twofold higher in the bottom 20% of peaks than in the upper 80% (genome structure correction (GSC)20 P value ,10216), consistent with previous observations2124. We speculate that low signal regions are either lower-affinity sites21 or indirect transcription-factor target regions associated through interactions with other factors (see also refs 25, 26). We organized all the information associated with each transcription factorincluding the ChIP-seq peaks, discovered motifs and associated histone modification patternsin FactorBook (http://www. factorbook.org; ref. 26), a public resource that will be updated as the project proceeds. DNase I hypersensitive sites and footprints Chromatin accessibility characterized by DNase I hypersensitivity is the hallmark of regulatory DNA regions27,28. We mapped 2.89 million unique, non-overlapping DNase I hypersensitive sites (DHSs) by DNase-seq in 125 cell types, the overwhelming majority of which lie distal to TSSs29. We also mapped 4.8 million sites across 25 cell types
Table 2 | Summary of ENCODE histone modifications and variants
Histone modification or variant Signal characteristics
that displayed reduced nucleosomal crosslinking by FAIRE, many of which coincide with DHSs. In addition, we used micrococcal nuclease to map nucleosome occupancy in GM12878 and K562 cells30. In tier 1 and tier 2 cell types, we identified a mean of 205,109 DHSs per cell type (at false discovery rate (FDR) 1%), encompassing an average of 1.0% of the genomic sequence in each cell type, and 3.9% in aggregate. On average, 98.5% of the occupancy sites of transcription factors mapped by ENCODE ChIP-seq (and, collectively, 94.4% of all 1.1 million transcription factor ChIP-seq peaks in K562 cells) lie within accessible chromatin defined by DNase I hotspots29. However, a small number of factors, most prominently heterochromatin-bound repressive complexes (for example, the TRIM28SETDB1ZNF274 complex31,32 encoded by the TRIM28, SETDB1 and ZNF274 genes), seem to occupy a significant fraction of nucleosomal sites. Using genomic DNase I footprinting33,34 on 41 cell types we identified 8.4 million distinct DNase I footprints (FDR 1%)25. Our de novo motif discovery on DNase I footprints recovered ,90% of known transcription factor motifs, together with hundreds of novel evolutionarily conserved motifs, many displaying highly cell-selective occupancy patterns similar to major developmental and tissue-specific regulators.
Regions of histone modification We assayed chromosomal locations for up to 12 histone modifications and variants in 46 cell types, including a complete matrix of eight modifications across tier 1 and tier 2. Because modification states may span multiple nucleosomes, which themselves can vary in position across cell populations, we used a continuous signal measure of histone modifications in downstream analysis, rather than calling regions (M. M. Hoffman et al., manuscript in preparation; see http://code. google.com/p/align2rawsignal/). For the strongest, peak-like histone modifications, we used MACS35 to characterize enriched sites. Table 2 describes the different histone modifications, their peak characteristics, and a summary of their known roles (reviewed in refs 3639). Our data show that global patterns of modification are highly variable across cell types, in accordance with changes in transcriptional activity. Consistent with previous studies40,41, we find that integration of the different histone modification information can be used systematically to assign functional attributes to genomic regions (see below). DNA methylation Methylation of cytosine, usually at CpG dinucleotides, is involved in epigenetic regulation of gene expression. Promoter methylation is typically associated with repression, whereas genic methylation correlates with transcriptional activity42. We used reduced representation bisulphite sequencing (RRBS) to profile DNA methylation quantitatively for an average of 1.2 million CpGs in each of 82 cell lines and tissues (8.6% of non-repetitive genomic CpGs), including CpGs in intergenic regions, proximal promoters and intragenic regions (gene bodies)43, although it should be noted that the RRBS method preferentially targets CpG-rich islands. We found that 96% of CpGs exhibited differential methylation in at least one cell type or tissue
Putative functions
H2A.Z H3K4me1 H3K4me2 H3K4me3 H3K9ac H3K9me1 H3K9me3 H3K27ac H3K27me3 H3K36me3 H3K79me2 H4K20me1
Peak Histone protein variant (H2A.Z) associated with regulatory elements with dynamic chromatin Peak/region Mark of regulatory elements associated with enhancers and other distal elements, but also enriched downstream of transcription starts Peak Mark of regulatory elements associated with promoters and enhancers Peak Mark of regulatory elements primarily associated with promoters/transcription starts Peak Mark of active regulatory elements with preference for promoters Region Preference for the 59 end of genes Peak/region Repressive mark associated with constitutive heterochromatin and repetitive elements Peak Mark of active regulatory elements; may distinguish active enhancers and promoters from their inactive counterparts Region Repressive mark established by polycomb complex activity associated with repressive domains and silent developmental genes Region Elongation mark associated with transcribed portions of genes, with preference for 39 regions after intron 1 Region Transcription-associated mark, with preference for 59 end of genes Region Preference for 59 end of genes
6 S E P T E M B E R 2 0 1 2 | VO L 4 8 9 | N AT U R E | 5 9
RESEARCH ARTICLE
assayed (K. Varley et al., personal communication), and levels of DNA methylation correlated with chromatin accessibility. The most variably methylated CpGs are found more often in gene bodies and intergenic regions, rather than in promoters and upstream regulatory regions. In addition, we identified an unexpected correspondence between unmethylated genic CpG islands and binding by P300, a histone acetyltransferase linked to enhancer activity44. Because RRBS is a sequence-based assay with single-base resolution, we were able to identify CpGs with allele-specific methylation consistent with genomic imprinting, and determined that these loci exhibit aberrant methylation in cancer cell lines (K. Varley et al., personal communication). Furthermore, we detected reproducible cytosine methylation outside CpG dinucleotides in adult tissues45, providing further support that this non-canonical methylation event may have important roles in human biology (K. Varley et al., personal communication).
Chromosome-interacting regions Physical interaction between distinct chromosome regions that can be separated by hundreds of kilobases is thought to be important in the regulation of gene expression46. We used two complementary chromosome conformation capture (3C)-based technologies to probe these long-range physical interactions. A 3C-carbon copy (5C) approach47,48 provided unbiased detection of long-range interactions with TSSs in a targeted 1% of the genome (the 44 ENCODE pilot regions) in four cell types (GM12878, K562, HeLa-S3 and H1 hESC)49. We discovered hundreds of statistically significant long-range interactions in each cell type after accounting for chromatin polymer behaviour and experimental variation. Pairs of interacting loci showed strong correlation between the gene expression level of the TSS and the presence of specific functional element classes such as enhancers. The average number of distal elements interacting with a TSS was 3.9, and the average number of TSSs interacting with a distal element was 2.5, indicating a complex network of interconnected chromatin. Such interwoven long-range architecture was also uncovered genome-wide using chromatin interaction analysis with paired-end tag sequencing (ChIA-PET)50 applied to identify interactions in chromatin enriched by RNA polymerase II (Pol II) ChIP from five cell types51. In K562 cells, we identified 127,417 promoter-centred chromatin interactions using ChIA-PET, 98% of which were intra-chromosomal. Whereas promoter regions of 2,324 genes were involved in single-gene enhancerpromoter interactions, those of 19,813 genes were involved in multi-gene interaction complexes spanning up to several megabases, including promoter promoter and enhancerpromoter interactions51. These analyses portray a complex landscape of long-range gene element connectivity across ranges of hundreds of kilobases to several megabases, including interactions among unrelated genes (Supplementary Fig. 1, section Y). Furthermore, in the 5C results, 5060% of longrange interactions occurred in only one of the four cell lines, indicative of a high degree of tissue specificity for geneelement connectivity49. Summary of ENCODE-identified elements Accounting for all these elements, a surprisingly large amount of the human genome, 80.4%, is covered by at least one ENCODE-identified element (detailed in Supplementary Table 1, section Q). The broadest element class represents the different RNA types, covering 62% of the genome (although the majority is inside of introns or near genes). Regions highly enriched for histone modifications form the next largest class (56.1%). Excluding RNA elements and broad histone elements, 44.2% of the genome is covered. Smaller proportions of the genome are occupied by regions of open chromatin (15.2%) or sites of transcription factor binding (8.1%), with 19.4% covered by at least one DHS or transcription factor ChIP-seq peak across all cell lines. Using our most conservative assessment, 8.5% of bases are covered by either a transcription-factor-binding-site motif (4.6%)
or a DHS footprint (5.7%). This, however, is still about 4.5-fold higher than the amount of protein-coding exons, and about twofold higher than the estimated amount of pan-mammalian constraint. Given that the ENCODE project did not assay all cell types, or all transcription factors, and in particular has sampled few specialized or developmentally restricted cell lineages, these proportions must be underestimates of the total amount of functional bases. However, many assays were performed on more than one cell type, allowing assessment of the rate of discovery of new elements. For both DHSs and CTCF-bound sites, the number of new elements initially increases rapidly with a steep gradient for the saturation curve and then slows with increasing number of cell types (Supplementary Figs 1 and 2, section R). With the current data, at the flattest part of the saturation curve each new cell type adds, on average, 9,500 DHS elements (across 106 cell types) and 500 CTCF-binding elements (across 49 cell types), representing 0.45% of the total element number. We modelled saturation for the DHSs and CTCF-binding sites using a Weibull distribution (r2 . 0.999) and predict saturation at approximately 4.1 million (standard error (s.e.) 5 108,000) and 185,100 (s.e. 5 18,020) sites, respectively, indicating that we have discovered around half of the estimated total DHSs. These estimates represent a lower bound, but reinforce the observation that there is more non-coding functional DNA than either coding sequence or mammalian evolutionarily constrained bases.
The impact of selection on functional elements From comparative genomic studies, at least 38% of bases are under purifying (negative) selection411, indicating that these bases may potentially be functional. We previously found that 60% of mammalian evolutionarily constrained bases were annotated in the ENCODE pilot project, but also observed that many functional elements lacked evidence of constraint2, a conclusion substantiated by others5254. The diversity and genome-wide occurrence of functional elements now identified provides an unprecedented opportunity to examine further the forces of negative selection on human functional sequences. We examined negative selection using two measures that highlight different periods of selection in the human genome. The first measure, inter-species, pan-mammalian constraint (GERP-based scores; 24 mammals8), addresses selection during mammalian evolution. The second measure is intra-species constraint estimated from the numbers of variants discovered in human populations using data from the 1000 Genomes project55, and covers selection over human evolution. In Fig. 1, we plot both these measures of constraint for different classes of identified functional elements, excluding features overlapping exons and promoters that are known to be constrained. Each graph also shows genomic background levels and measures of coding-gene constraint for comparison. Because we plot human population diversity on an inverted scale, elements that are more constrained by negative selection will tend to lie in the upper and right-hand regions of the plot. For DNase I elements (Fig. 1b) and bound motifs (Fig. 1c), most sets of elements show enrichment in pan-mammalian constraint and decreased human population diversity, although for some cell types the DNase I sites do not seem overall to be subject to pan-mammalian constraint. Bound transcription factor motifs have a natural control from the set of transcription factor motifs with equal sequence potential for binding but without binding evidence from ChIP-seq experimentsin all cases, the bound motifs show both more mammalian constraint and higher suppression of human diversity. Consistent with previous findings, we do not observe genome-wide evidence for pan-mammalian selection of novel RNA sequences (Fig. 1d). There are also a large number of elements without mammalian constraint, between 17% and 90% for transcription-factor-binding regions as well as DHSs and FAIRE regions. Previous studies could not determine whether these sequences are either biochemically active, but with little overall impact on the organism, or under lineagespecific selection. By isolating sequences preferentially inserted into
ARTICLE RESEARCH
b
3 104
2 104
Human and mammalian constraint at TSS-distal non-exonic DNase peaks
Human and mammalian constraint at TSS-distal non-exonic ENCODE elements
Human and mammalian constraint at regulatory motifs, bound and unbound by their cognate factors
Human diversity (inverted scale)
AP2 MYC EBF
4 104
2 104
4 104
EGR1
RXRA YY1 ETS CTCF TCF4 HNF4Irf U AP1 STAT MAFNRSF GR Pou2f2 Foxa MEF2 CEBPB GATA NFB SRF
5 104
8 104 6 104
G
SP1
6 104
4 104
7 104
1.0
0.5
0.0
0.5
6 104
Mammalian conservation
IN G IG
1 103
Motifs bound by their factor(s) in > 1 experiment Motifs never bound by their factor(s)
1.0
0.5
0.0
0.5
1.0
3 104
8 104 1.0
Human and mammalian constraint at novel (non-GENCODE-annotated) long intronic or intergenic TSS-distal transcriptionally active regions
0.5
0.0
0.5
1.0
1.5
2.0
U
f
5
Glucocorticoid receptor transcription factor
4 104
e
0.30
Primate-specific TSS-distal non-exonic regions GERP conservation score
4 3 2 1 0
r = 0.956 P = 2.67 108
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 (+1) (+2) (+3) (+4) (+5) (+6) (+7) (+8) (+9) (+10)
5 104
IN
Fraction
0.20 Non-ENCODE ENCODE-annotated 0.10
6 104
IG
7 104
0.00 1.0 0.5 0.0 0.5
DAF
(10) (9) (8) (7) (6) (5) (4) (3) (2) (1)
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
Motif position
Figure 1 | Impact of selection on ENCODE functional elements in mammals and human populations. a, Levels of pan-mammalian constraint (mean GERP score; 24 mammals8, x axis) compared to diversity, a measure of negative selection in the human population (mean expected heterozygosity, inverted scale, y axis) for ENCODE data sets. Each point is an average for a single data set. The top-right corners have the strongest evolutionary constraint and lowest diversity. Coding (C), UTR (U), genomic (G), intergenic (IG) and intronic (IN) averages are shown as filled squares. In each case the vertical and horizontal cross hairs show representative levels for the neutral expectation for mammalian conservation and human population diversity, respectively. The spread over all non-exonic ENCODE elements greater than 2.5 kb from TSSs is shown. The inner dashed box indicates that parts of the plot have been magnified for the surrounding outer panels, although the scales in the outer plots provide the exact regions and dimensions magnified. The spread for DHS sites (b) and RNA elements (d) is shown in the plots on the left. RNA elements
are either long novel intronic (dark green) or long intergenic (light green) RNAs. The horizontal cross hairs are colour-coded to the relevant data set in d. c, Spread of transcription factor motif instances either in regions bound by the transcription factor (orange points) or in the corresponding unbound motif matches in grey, with bound and unbound points connected with an arrow in each case showing that bound sites are generally more constrained and less diverse. e, Derived allele frequency spectrum for primate-specific elements, with variations outside ENCODE elements in black and variations covered by ENCODE elements in red. The increase in low-frequency alleles compared to background is indicative of negative selection occurring in the set of variants annotated by the ENCODE data. f, Aggregation of mammalian constraint scores over the glucocorticoid receptor (GR) transcription factor motif in bound sites, showing the expected correlation with the information content of bases in the motif. An interactive version of this figure is available in the online version of the paper.
the primate lineage, which is only feasible given the genome-wide scale of this data, we are able to examine this issue specifically. Most primatespecific sequence is due to retrotransposon activity, but an appreciable proportion is non-repetitive primate-specific sequence. Of 104,343,413 primate-specific bases (excluding repetitive elements), 67,769,372 (65%) are found within ENCODE-identified elements. Examination of 227,688 variants segregating in these primate-specific regions revealed that all classes of elements (RNA and regulatory) show depressed derived allele frequencies, consistent with recent negative selection occurring in at least some of these regions (Fig. 1e). An alternative approach examining sequences that are not clearly under panmammalian constraint showed a similar result (L. Ward and M. Kellis, manuscript submitted). This indicates that an appreciable proportion of the unconstrained elements are lineage-specific elements required for organismal function, consistent with long-standing views of recent evolution56, and the remainder are probably neutral elements2 that are not currently under selection but may still affect cellular or larger scale phenotypes without an effect on fitness.
The binding patterns of transcription factors are not uniform, and we can correlate both inter- and intra-species measures of negative selection with the overall information content of motif positions. The selection on some motif positions is as high as protein-coding exons (Fig. 1f; L. Ward and M. Kellis, manuscript submitted). These aggregate measures across motifs show that the binding preferences found in the population of sites are also relevant to the per-site behaviour. By developing a per-site metric of population effect on bound motifs, we found that highly constrained bound instances across mammals are able to buffer the impact of individual variation57.
ENCODE data integration with known genomic features

Promoter-anchored integration Many of the ENCODE assays directly or indirectly provide information about the action of promoters. Focusing on the TSSs of proteincoding transcripts, we investigated the relationships between different ENCODE assays, in particular testing the hypothesis that RNA expression (output) can be effectively predicted from patterns of
RESEARCH ARTICLE
chromatin modification or transcription factor binding (input). Consistent with previous reports58, we observe two relatively distinct types of promoter: (1) broad, mainly (C1G)-rich, TATA-less promoters; and (2) narrow, TATA-box-containing promoters. These promoters have distinct patterns of histone modifications, and transcription-factor-binding sites are selectively enriched in each class (Supplementary Fig. 1, section Z). We developed predictive models to explore the interaction between histone modifications and measures of transcription at promoters, distinguishing between modifications known to be added as a consequence of transcription (such as H3K36me3 and H3K79me2) and other categories of histone marks59. In our analyses, the best models had two components: an initial classification component (on/off) and a second quantitative model component. Our models showed that activating acetylation marks (H3K27ac and H3K9ac) are roughly as informative as activating methylation marks (H3K4me3 and H3K4me2) (Fig. 2a). Although repressive marks, such as H3K27me3
Classification (mean decreased Gini)
or H3K9me3, show negative correlation both individually and in the model, removing these marks produces only a small reduction in model performance. However, for a subset of promoters in each cell line, repressive histone marks (H3K27me3 or H3K9me3) must be used to predict their expression accurately. We also examined the interplay between the H3K79me2 and H3K36me3 marks, both of which mark gene bodies, probably reflecting recruitment of modification enzymes by polymerase isoforms. As described previously, H3K79me2 occurs preferentially at the 59 ends of gene bodies and H3K36me3 occurs more 39, and our analyses support the previous model in which the H3K79me2 to H3K36me3 transition occurs at the first 39 splice site60. Few previous studies have attempted to build qualitative or quantitative models of transcription genome-wide from transcription factor levels because of the paucity of documented transcriptionfactor-binding regions and the lack of coordination around a single cell line. We thus examined the predictive capacity of transcriptionfactor-binding signals for the expression levels of promoters (Fig. 2b).
Relative importance of variables
CAGE poly(A)+ K562 whole cell

1,000 800 600 400 200 0
Pearsons r = 0.9 RMSE = 1.9 10

Measured expression (log2)
Classification: AUC = 0.95 Regression: r = 0.78 (RMSE = 2.23)
Regression (mean decreased Gini)
15,000
10,000
5,000
5 4 2 0 2 4 Predicted expression (log2) 6 8
CAGE poly(A)+ K562 whole cell Classification (mean decreased Gini) Pearsons r = 0.81 RMSE = 2.57 10
Classification: AUC = 0.89 Regression: r = 0.62 (RMSE = 3.06)
80 60 40 20 0
Measured expression (log2)
5 Regression (mean decreased Gini)

3,500 3,000 2,500 2,000 1,500 1,000 500
5 4 2 0 2 4 Predicted expression (log2) 6 8
Figure 2 | Modelling transcription levels from histone modification and transcription-factor-binding patterns. a, b, Correlative models between either histone modifications or transcription factors, respectively, and RNA production as measured by CAGE tag density at TSSs in K562 cells. In each case the scatter plot shows the output of the correlation models (x axis) compared to observed values (y axis). The bar graphs show the most important histone
YY M 1 Y EL C F E2 1 F JU 4 ET N RE S1 S M T G XI AB 1 P M A A FO X N S F EGYB R E2 1 ZN F F 6 TH 26 AP 3 S 1 N P2 RF TA 1 L ZBNFY1 TB A G 7A AT A SP 2 SI I1 X5 U FOSF2 SL S 1 G P1 AT BCA1 L BCUSF 3 LA 1 JU F1 ZB N TB D N 33 F M E2 AF K C SR EB F PB A N TF R 3 ZN 2C F2 2 74
YY M 1 Y M C RE AX ST ET S E2 1 F EL 4 EG F1 R1 TA L JU 1 E2 N BC F LA 6 F M 1 XI 1 G SP ZBABP1 TB A B 7A TH CL AP 3 S 1 G RF AT U A2 S N F2 FOFYA S M L1 ZB AF TB K 3 FO 3 N S ZN RF F2 1 U 63 G SF1 AT A SP 1 AT 2 C F E 3 N BPB R2 JU C2 N SP D N I1 FY B ZN SIX F2 5 N 74 FE 2
3K 9 3K ac 4 3K me 36 3 H me3 3K 27 DN ac H ase 3K I 4m e H 2 N H3K 2A. or Z m 27 al m ize e3 d H Cp 3K G 4 H me 3K 1 9 H me 3K 1 H 9m 4K e 20 3 m e C 1 on tro l H H
79 H 3K
e2
H 3K 9 3K ac 4 H me 3K 3 4m DN e2 a H se 3 I H K27 3K a 79 c m e2 H H2A 3K 36 .Z H 3K me 27 3 N H3 me or m K4 3 al m ize e1 d H Cp 3K G 9 H me 3K 1 H 9m 4K e 20 3 m e C 1 on tro l
Relative importance of variables
modifications (a) or transcription factors (b) in both the initial classification phase (top bar graph) or the quantitative regression phase (bottom bar graph), with larger values indicating increasing importance of the variable in the model. Further analysis of other cell lines and RNA measurement types is reported elsewhere59,79. AUC, area under curve; Gini, Gini coefficient; RMSE, root mean square error.
ARTICLE RESEARCH
In contrast to the profiles of histone modifications, most transcription factors show enriched binding signals in a narrow DNA region near the TSS, with relatively higher binding signals in promoters with higher CpG content. Most of this correlation could be recapitulated by looking at the aggregate binding of transcription factors without specific transcription factor terms. Together, these correlation models indicate both that a limited set of chromatin marks are sufficient to explain transcription and that a variety of transcription factors might have broad roles in general transcription levels across many genes. It is important to note that this is an inherently observational study of correlation patterns, and is consistent with a variety of mechanistic models with different causal links between the chromatin, transcription factor and RNA assays. However, it does indicate that there is enough information present at the promoter regions of genes to explain most of the variation in RNA expression. We developed predictive models similar to those used to model transcriptional activity to explore the relationship between levels of histone modification and inclusion of exons in alternately spliced transcripts. Even accounting for expression level, H3K36me3 has a positive contribution to exon inclusion, whereas H3K79me2 has a negative contribution (H. Tilgner et al., manuscript in preparation). By monitoring the RNA populations in the subcellular fractions of K562 cells, we found that essentially all splicing is co-transcriptional61, further supporting a link between chromatin structure and splicing.
Transcription-factor-binding site-anchored integration Transcription-factor-binding sites provide a natural focus around which to explore chromatin properties. Transcription factors are often multifunctional and can bind a variety of genomic loci with different combinations and patterns of chromatin marks and nucleosome organization. Hence, rather than averaging chromatin mark profiles across all binding sites of a transcription factor, we developed a clustering procedure, termed the Clustered Aggregation Tool (CAGT), to identify subsets of binding sites sharing similar but distinct patterns of chromatin mark signal magnitude, shape and hidden directionality30. For example, the average profile of the repressive histone mark H3K27me3 over all 55,782 CTCF-binding sites in H1 hESCs shows poor signal enrichment (Fig. 3a). However, after grouping profiles by signal magnitude we found a subset of 9,840 (17.6%) CTCF-binding sites that exhibit significant flanking H3K27me3 signal. Shape and orientation analysis further revealed that the predominant signal profile for H3K27me3 around CTCF peak summits is asymmetric, consistent with a boundary role for some CTCF sites between active and polycomb-silenced domains. Further examples are provided in Supplementary Figs 5 and 6 of section E. For TAF1, predominantly found near TSSs, the asymmetric sites are orientated with the direction of transcription. However, for distal sites, such as those bound by GATA1 and CTCF, we also observed a high proportion of asymmetric histone patterns, although independent of motif directionality. In fact, all transcription-factor-binding data sets in all cell lines show predominantly asymmetric patterns (asymmetry ratio .0.6) for all chromatin marks but not for DNase I signal (Fig. 3b). This indicates that most transcription-factor-bound chromatin events correlate with structured, directional patterns of histone modifications, and that promoter directionality is not the only source of orientation at these sites. We also examined nucleosome occupancy relative to the symmetry properties of chromatin marks around transcription-factor-binding sites. Around TSSs, there is usually strong asymmetric nucleosome occupancy, often accounting for most of the histone modification signal (for instance, see Supplementary Fig. 4, section E). However, away from TSSs, there is far less concordance. For example, CTCFbinding sites typically show arrays of well-positioned nucleosomes on either side of the peak summit (Supplementary Fig. 1, section E)62. Where the flanking chromatin mark signal is high, the signals are often asymmetric, indicating differential marking with histone modifications (Supplementary Figs 2 and 3, section E). Thus, we
a H3K27me3 at CTCF in H1 hESC (TSS-proximal/distal transcription factor)
All (55,782) 35 30 25 20 15 10 5 0 High (17.64%) 8 Pattern 3 (1.59%) 40 30 20 10 0 Pattern 5 (0.56%) 20 15 15 10 5 0 400 0 400 10 5 0 400 0 400 10 5 0 400 0 400 15 Pattern 6 (0.55%) Pattern 1 (10.14%) 15 10 5 0 Pattern 4 (1.06%) Pattern 2 (2.88%)
15 10 5 0
Signal
Signal
Low (82.36%)
15 10 5 0
6 4 2 0
Distance from summit
Distance from summit
b
1.0 0.8 0.6 0.4 0.2 0.0
DNase
H2A.Z
H3K27ac
H3K27me3
Cumulative distribution
1.0 0.8 0.6 0.4 0.2 0.0 1.0 0.8 0.6 0.4 0.2 0.0 1.0 0.8 0.6 0.4 0.2 0.0
H3K36me3
H3K4me1
H3K4me2
H3K4me3
H3K79me2
H3K9ac
H3K9me1
H3K9me3
H4K20me1
Nucleosome CTCF, CTCFL, RAD21, SMC3, ZNF143
0.2 0.4 0.6 0.8 1.0 0.2 0.4 0.6 0.8 1.0 0.2 0.4 0.6 0.8 1.0 0.2 0.4 0.6 0.8 1.0
Asymmetry ratio
Figure 3 | Patterns and asymmetry of chromatin modification at transcription-factor-binding sites. a, Results of clustered aggregation of H3K27me3 modification signal around CTCF-binding sites (a multifunctional protein involved with chromatin structure). The first three plots (left column) show the signal behaviour of the histone modification over all sites (top) and then split into the high and low signal components. The solid lines show the mean signal distribution by relative position with the blue shaded area delimiting the tenth and ninetieth percentile range. The high signal component is then decomposed further into six different shape classes on the right (see ref. 30 for details). The shape decomposition process is strand aware. b, Summary of shape asymmetry for DNase I, nucleosome and histone modification signals by plotting an asymmetry ratio for each signal over all transcription-factorbinding sites. All histone modifications measured in this study show predominantly asymmetric patterns at transcription-factor-binding sites. An interactive version of this figure is available in the online version of the paper.
confirm on a genome-wide scale that transcription factors can form barriers around which nucleosomes and histone modifications are arranged in a variety of configurations6265. This is explored in further detail in refs 25, 26 and 30.
Transcription factor co-associations Transcription-factor-binding regions are nonrandomly distributed across the genome, with respect to both other features (for example, promoters) and other transcription-factor-binding regions. Within the
RESEARCH ARTICLE
TAF1 YY1 TBP E2F4 E2F6 ELF1 MAX POLR2A HMGN3 ZBTB7A CCNT2 EGR1 ETS1 SIN3A HDAC2 GABPA MXI1 MYC CHD2 IRF1 GTF2F1 THAP1 SP2 REST NRF1 USF1 FOS SP1 SRF SPI1 SIX5 CTCF RAD21 SMC3 CTCFL ZNF263 BCLAF1 TAF7 RDBP ZBTB33 BCL3 ATF3 USF2 NFE2 SETDB1 TRIM28 ZNF274 NR2C2 GATA1 GATA2 TAL1 EP300 SMARCA4 SMARCB1 SIRT6 JUNB JUND JUN FOSL1 MAFK CEBPB HDAC8 STAT1 STAT2 BDP1 POLR3A BRF1 GTF3C2 BRF2
a
TAF1 TBP YY1 ELF1 MAX E2F4 E2F6 IRF1 EGR1 ZBTB7A ETS1 SIN3A CCNT2 HMGN3 HDAC2 GABPA CHD2 POLR2A GTF2F1 MXI1 MYC THAP1 SP1 SP2 NRF1 REST SIX5 SRF SPI1 RAD21 SMC3 CTCF CTCFL ZNF263 BCLAF1 TAF7 RDBP ZBTB33 BCL3 ATF3 USF2 USF1 NFE2 GATA1 GATA2 TAL1 EP300 SMARCA4 SMARCB1 SIRT6 JUNB JUND JUN FOSL1 FOS MAFK CEBPB HDAC8 SETDB1 TRIM28 NR2C2 ZNF274 STAT1 STAT2 BDP1 POLR3A BRF1 GTF3C2 BRF2
Whole genome
b
Intergenic regions Promotor proximal regions
HDAC2 GABPA
CHD2
MXI1
MYC
JUNB JUND JUN FOSL1
HDAC2 GABPA CHD2 POLR2A GTF2F1 MXI1 MYC HDAC2 GABPA CHD2 MXI1 MYC
FOSL1
JUND
JUNB
JUN
HDAC2 NANOG EP300 SP1
JUNB JUND JUN FOSL1
HDAC2 GABPA CHD2 POLR2A GTF2F1 MXI1 MYC
FOSL1
JUND
JUNB
JUN
HDAC2 NANOG EP300 SP1
A (K562)
B (K562)
(H1 hESC)
Confidence
High
Medium
Low
0 10 20 30 40 50 60 70 80 90 100
Degree of co-association (z-score)
Figure 4 | Co-association between transcription factors. a, Significant coassociations of transcription factor pairs using the GSC statistic across the entire genome in K562 cells. The colour strength represents the extent of association (from red (strongest), orange, to yellow (weakest)), whereas the depth of colour represents the fit to the GSC20 model (where white indicates that the statistical model is not appropriate) as indicated by the key. Most transcription factors have a nonrandom association to other transcription factors, and these associations are dependent on the genomic context, meaning that once the genome is separated into promoter proximal and distal regions, the overall levels of co-association
decrease, but more specific relationships are uncovered. b, Three classes of behaviour are shown. The first column shows a set of associations for which strength is independent of location in promoter and distal regions, whereas the second column shows a set of transcription factors that have stronger associations in promoter-proximal regions. Both of these examples are from data in K562 cells and are highlighted on the genome-wide co-association matrix (a) by the labelled boxes A and B, respectively. The third column shows a set of transcription factors that show stronger association in distal regions (in the H1 hESC line). An interactive version of this figure is available in the online version of the paper.
tier 1 and 2 cell lines, we found 3,307 pairs of statistically co-associated factors (P ,1 3 10216, GSC) involving 114 out of a possible 117 factors (97%) (Fig. 4a). These include expected associations, such as Jun and
Table 3 | Summary of the combined state types
Label Description
Fos, and some less expected novel associations, such as TCF7L2 with HNF4-a and FOXA2 (ref. 66; a full listing is given in Supplementary Table 1, section F). When one considers promoter and intergenic
Details*
Colour
CTCF
PF R
TSS
T WE
Sites of CTCF signal lacking histone modifications, often associated with open chromatin. Many probably have a function in insulator assays, but because of the multifunctional nature of CTCF, we are conservative in our description. Also enriched for the cohesin components RAD21 and SMC3; CTCF is known to recruit the cohesin complex. Predicted enhancer Regions of open chromatin associated with H3K4me1 signal. Enriched for other enhancerassociated marks, including transcription factors known to act at enhancers. In enhancer assays, many of these (.50%) function as enhancers. A more conservative alternative would be cisregulatory regions. Enriched for sites for the proteins encoded by EP300, FOS, FOSL1, GATA2, HDAC8, JUNB, JUND, NFE2, SMARCA4, SMARCB1, SIRT6 and TAL1 genes in K562 cells. Have nuclear and whole-cell RNA signal, particularly poly(A)2 fraction. Predicted promoter flanking region Regions that generally surround TSS segments (see below). Predicted repressed or low-activity region This is a merged state that includes H3K27me3 polycomb-enriched regions, along with regions that are silent in terms of observed signal for the input assays to the segmentations (low or no signal). They may have other signals (for example, RNA, not in the segmentation input data). Enriched for sites for the proteins encoded by REST and some other factors (for example, proteins encoded by BRF2, CEBPB, MAFK, TRIM28, ZNF274 and SETDB1 genes in K562 cells). Predicted promoter region including TSS Found close to or overlapping GENCODE TSS sites. High precision/recall for TSSs. Enriched for H3K4me3. Sites of open chromatin. Enriched for transcription factors known to act close to promoters and polymerases Pol II and Pol III. Short RNAs are most enriched in these segments. Predicted transcribed region Overlap gene bodies with H3K36me3 transcriptional elongation signal. Enriched for phosphorylated form of Pol II signal (elongating polymerase) and poly(A)1 RNA, especially cytoplasmic. Predicted weak enhancer or open Similar to the E state, but weaker signals and weaker enrichments. chromatin cis-regulatory element
CTCF-enriched element
Turquoise
Orange
Light red Grey
Bright red
Dark green Yellow
* Where specific enrichments or overlaps are identified, these are derived from analysis in GM12878 and/or K562 cells where the data for comparison is richest. The colours indicated are used in Figs 5 and 7 and in display of these tracks from the ENCODE data hub.
NANOG
HDAC2
EP300
SP1
NANOG
HDAC2
EP300
SP1
ARTICLE RESEARCH
a
Segmentations
Chr22:
GENCODE v7 genes ChromHMM Segway GM12878 combined
R T WE E CTCF PF TSS
46100000
46200000
46300000
46400000
46500000
46600000
46700000
UW DNase Open chrom DNase FAIRE H3K4me1
Signal tracks
H3K4me2 H3K4me3 H3K9ac H3K27ac H3K27me3 H3K36me3 H4K20me1 CTCF Pol II Input control
GM12878 segment overlaps with transcription factors (obs./exp. coverage)

POU2F2 SLC22A2 TCF12 EGR1 SRF EBF1 BCL11A NFKB1 IRF4 PAX5 PBX3 USF1 ZEB1 RFX5 NFE2 ZBTB33 SPI1 EBF1 BATF MEF2A MEF2C YY1 USF2 JUND ZZZ3 RES T BCL3 NR2C2 YY1 ATF3 RXRA ELF1 MAX SP1 BCLAF1 SIX5 ETS1 SIN3A TBP GABPA CTCF RAD21 ZNF143 RAD21 SMC3 WRNIP1 EP300 STAT3 POLR3A NRF1 CHD2 FOS POLR2A BRCA1 TAF1
GM12878 overlaps with RNAs (obs./exp.)

Long RNA whole cell A Long RNA nucleus A+ Long RNA nucleus A Long RNA whole cell A+
Enriched (log2 = 6.5)
Long RNA cytoplasm A+ Long RNA cytoplasm A
Enriched (log2 = 5.8)
(log2 = 0)
Short RNA cytosol Short RNA whole cell Short RNA nucleus
(log2 = 0)
WE
CTCF
PF
CTCF
Segment class
TSS
Segment class
Cell type in each T segment
250,000 8,000
c
Segment count
25,000
250,000
Per cent methylated
Cell type in each CTCF segment

40,000
Cell type in each E segment
Cell type in each TSS segment
Cell type in each R segment
d
100
Distribution of methylation over GM12878 segment labels
60
100,000
10,000
100,000
20,000
4,000
20 0
CTCF
Mean cell type in segment
Figure 5 | Integration of ENCODE data by genome-wide segmentation. a, Illustrative region with the two segmentation methods (ChromHMM and Segway) in a dense view and the combined segmentation expanded to show each state in GM12878 cells, beneath a compressed view of the GENCODE gene annotations. Note that at this level of zoom and genome browser resolution, some segments appear to overlap although they do not. Segmentation classes are named and coloured according to the scheme in Table 3. Beneath the segmentations are shown each of the normalized signals that were used as the input data for the segmentations. Open chromatin signals from DNase-seq from the University of Washington group (UW DNase) or the ENCODE open chromatin group (Openchrom DNase) and FAIRE assays are shown in blue; signal from histone modification ChIP-seq in red; and transcription factor ChIP-seq signal for Pol II and CTCF in green. The mauve
ChIP-seq control signal (input control) at the bottom was also included as an input to the segmentation. b, Association of selected transcription factor (left) and RNA (right) elements in the combined segmentation states (x axis) expressed as an observed/expected ratio (obs./exp.) for each combination of transcription factor or RNA element and segmentation class using the heatmap scale shown in the key besides each heat map. c, Variability of states between cell lines, showing the distribution of occurrences of the state in the six cell lines at specific genome locations: from unique to one cell line to ubiquitous in all six cell lines for five states (CTCF, E, T, TSS and R). d, Distribution of methylation level at individual sites from RRBS analysis in GM12878 cells across the different states, showing the expected hypomethylation at TSSs and hypermethylation of genes bodies (T state) and repressed (R) regions.
TSS
WE
PF
1 2 3 4 5 6
1 2 3 4 5 6
1 2 3 4 5 6
1 2 3 4 5 6
1 2 3 4 5 6
TSS
WE
PF
Depleted (log2 = 10.0)
Depleted (log2 = 3.8)
RESEARCH ARTICLE
regions separately, this changes to 3,201 pairs (116 factors, 99%) for promoters and 1,564 pairs (108 factors, 92%) for intergenic regions, with some associations more specific to these genomic contexts (for example, the cluster of HDAC2, GABPA, CHD2, GTF2F1, MXI1 and MYC in promoter regions and SP1, EP300, HDAC2 and NANOG in intergenic regions (Fig. 4b)). These general and context-dependent associations lead to a network representation of the co-binding with many interesting properties, explored in refs 19, 25 and 26. In addition, we also identified a set of regions bound by multiple factors representing high occupancy of transcription factor (HOT) regions67. states also have an excess of RNA elements without poly(A) tails and methyl-cap RNA, as assayed by CAGE sequences, compared to matched intergenic controls, indicating a specific transcriptional mode associated with active enhancers70. Transcription factors also showed distinct distributions across the segments (Fig. 5b). A striking pattern is the concentration of transcription factors in the TSSassociated state. The enhancers contain a different set of transcription factors. For example, in K562 cells, the E state is enriched for binding by the proteins encoded by the EP300, FOS, FOSL1, GATA2, HDAC8, JUNB, JUND, NFE2, SMARCA4, SMARCB1, SIRT6 and TAL1 genes. We tested a subset of these predicted enhancers in both mouse and fish transgenic models (examples in Fig. 6), with over half of the elements showing activity, often in the corresponding tissue type. The segmentation provides a linear determination of functional state across the genome, but not an association of particular distal regions with genes. By using the variation of DNase I signal across cell lines, 39% of E (enhancer associated) states could be linked to a proposed regulated gene29 concordant with physical proximity patterns determined by 5C49 or ChIA-PET. To provide a fine-grained regional classification, we turned to a self organizing map (SOM) to cluster genome segmentation regions based on their assay signal characteristics (Fig. 7). The segmentation regions were initially randomly assigned to a 1,350-state map in a twodimensional toroidal space (Fig. 7a). This map can be visualized as a two-dimensional rectangular plane onto which the various signal distributions can be plotted. For instance, the rectangle at the bottom left of Fig. 7a shows the distribution of the genome in the initial randomized map. The SOM was then trained using the twelve different ChIP-seq and DNase-seq assays in the six cell types previously analysed in the large-scale segmentations (that is, over 72-dimensional space). After training, the SOM clustering was again visualized in two dimensions, now showing the organized distribution of genome segments (lower right of panel, Fig. 7a). Individual data sets associated with the genome segments in each SOM map unit (hexagonal cells) can then be visualized in the same framework to learn how each additional kind of data is distributed on the chromatin state map. Figure 7b shows CAGE/TSS expression data overlaid on the randomly initialized (left) and trained map (right) panels. In this way the trained SOM highlighted cell-type-specific TSS clusters (bottom panels of Fig. 7b), indicating that there are sets of tissue-specific TSSs that are distinguished from each other by subtle combinations of ENCODE
a b
Genome-wide integration
To identify functional regions genome-wide, we next integrated elements independent of genomic landmarks using either discriminative training methods, where a subset of known elements of a particular class were used to train a model that was then used to discover more instances of this class, or using methods in which only data from ENCODE assays were used without explicit knowledge of any annotation. For discriminative training, we used a three-step process to predict potential enhancers, described in Supplementary Information and ref. 67. Two alternative discriminative models converged on a set of ,13,000 putative enhancers in K562 cells67. In the second approach, two methodologically distinct unbiased approaches (see refs 40, 68 and M. M. Hoffman et al., manuscript in preparation) converged on a concordant set of histone modification and chromatin-accessibility patterns that can be used to segment the genome in each of the tier 1 and tier 2 cell lines, although the individual loci in each state in each cell line are different. With the exception of RNA polymerase II and CTCF, the addition of transcription factor data did not substantially alter these patterns. At this stage, we deliberately excluded RNA and methylation assays, reserving these data as a means to validate the segmentations. Our integration of the two segmentation methods (M. M. Hoffman et al., manuscript in preparation) established a consensus set of seven major classes of genome states, described in Table 3. The standard view of active promoters, with a distinct core promoter region (TSS and PF states), leading to active gene bodies (T, transcribed state), is rediscovered in this model (Fig. 5a, b). There are three active distal states. We tentatively labelled two as enhancers (predicted enhancers, E, and predicted weak enhancers, WE) due to their occurrence in regions of open chromatin with high H3K4me1, although they differ in the levels of marks such as H3K27ac, currently thought to distinguish active from inactive enhancers. The other active state (CTCF) has high CTCF binding and includes sequences that function as insulators in a transfection assay. The remaining repressed state (R) summarizes sequences split between different classes of actively repressed or inactive, quiescent chromatin. We found that the CTCF-binding-associated state is relatively invariant across cell types, with individual regions frequently occupying the CTCF state across all six cell types (Fig. 5c). Conversely, the E and T states have substantial cell-specific behaviour, whereas the TSS state has a bimodal behaviour with similar numbers of cell-invariant and cell-specific occurrences. It is important to note that the consensus summary classes do not capture all the detail discovered in the individual segmentations containing more states. The distribution of RNA species across segments is quite distinct, indicating that underlying biological activities are captured in the segmentation. Polyadenylated RNA is heavily enriched in gene bodies. Around promoters, there are short RNA species previously identified as promoter-associated short RNAs (Fig. 5b)16,69. Similarly, DNA methylation shows marked distinctions between segments, recapitulating the known biology of predominantly unmethylated active promoters (TSS states) followed by methylated gene bodies42 (T state, Fig. 5d). The two enhancer-enriched states show distinct patterns of DNA methylation, with the less active enhancer state (by H3K27ac/H3K4me1 levels) showing higher methylation. These
Figure 6 | Experimental characterization of segmentations. Randomly sampled E state segments (see Table 3) from the K562 segmentation were cloned for mouse- and fish-based transgenic enhancer assays. a, Representative LacZ-stained transgenic embryonic day (E)11.5 mouse embryo obtained with construct hs2065 (EN167, chr10: 4605288246055670, GRCh37). Highly reproducible staining in the blood vessels was observed in 9 out of 9 embryos resulting from independent transgenic integration events. b, Representative green fluorescent protein reporter transgenic medaka fish obtained from a construct with a basal hsp70 promoter on meganuclease-based transfection. Reproducible transgenic expression in the circulating nucleated blood cells and the endothelial cell walls was seen in 81 out of 100 transgenic tests of this construct.
ARTICLE RESEARCH
a
72 signal/cell line measurements at each state
Genome segments along chromosome Initialize map with genome segments at random Train map using 72 signal/cell line combinations
Empty map
Randomly initialized map visualized by DNA content

Transform to planar surface for viewing
Trained map
Transform to planar surface for viewing
0 4.8 4.2 3.6 3.0 2.4 1.8 1.2 0.6 5.0 4.5 4.0 3.5 3.0 2.5 2.0 1.5 0 44 1.6 1.4 1.2 1.0 0.8 0.6 0.4 0.2 0.0 0 44
Nucleotide distribution (log10)

29
29
b
CAGE/GENCODE expressed TSS
0 0.090 0.075 0.060 0.045 0.030 0.015 29 0 44
29
All expressed TSSs
TSS H1 hESC only
TSS HepG2 only
c
GO: immune response
0 32 28 24 20 16 12 8 4 0 0 44
GO: sequence-specific transcription factor activity

0 140 120 100 80 60 40 20 0 0 44
All significant GO terms/unit

0 40 35 30 25 20 15 10 5 0 0 44
29
29
29
H3K27me3 high in ES cell H3K27me3 high in HUVEC Primarily neuronal transcription factors
H3K27me3 high in ES cell H3K27me3 low in HUVEC Body patterning transcription factors
Scale Chr11: 20160000
20 kb
20180000 20200000
Scale Chr16:
20 kb
55350000 55370000 55390000
K562 GM12878 H1 hESC HeLa-S3 HepG2 HUVEC

DBX1
IRX6
Figure 7 | High-resolution segmentation of ENCODE data by self-organizing maps (SOM). ac, The training of the SOM (a) and analysis of the results (b, c) are shown. Initially we arbitrarily placed genomic segments from the ChromHMM segmentation on to the toroidal map surface, although the SOM does not use the ChromHMM state assignments (a). We then trained the map using the signal of the 12 different ChIP-seq and DNase-seq assays in the six cell types analysed. Each unit of the SOM is represented here by a hexagonal cell in a planar two-dimensional view of the toroidal map. Curved arrows indicate that traversing the edges of two dimensional view leads back to the opposite edge. The resulting map can be overlaid with any class of ENCODE or other data to view the distribution of that data within this high-resolution segmentation. In panel a the distributions of genome bases across the untrained and trained map (left and right, respectively) are shown using heat-map colours for log10 values. b, The distribution of TSSs from CAGE experiments of GENCODE annotation on the planar representations of either the initial random organization (left) or the final trained SOM (right) using heat maps coloured according to the accompanying scales. The bottom half of b expands the different distributions in the SOM for all expressed TSSs (left) or TSSs specifically expressed in two example cell lines, H1 hESC (centre) and HepG2 (right). c, The association of Gene Ontology (GO) terms on the same representation of the same trained SOM. We assigned genes that are within 20 kb of a genomic segment in a SOM unit to that unit, and then associated this set of genes with GO terms using a hypergeometric distribution after correcting for multiple testing. Map units that are significantly associated to GO terms are coloured green, with increasing strength of colour reflecting increasing numbers of genes significantly associated with the GO terms for either immune response (left) or sequence-specific transcription factor activity (centre). In each case, specific SOM units show association with these terms. The right-hand panel shows the distribution on the same SOM of all significantly associated GO terms, now colouring by GO term count per SOM unit. For sequence-specific transcription factor activity, two example genomic regions are extracted at the bottom of panel c from neighbouring SOM units. These are regions around the DBX1 (from SOM unit 26,31, left panel) and IRX6 (SOM unit 27,30, right panel) genes, respectively, along with their H3K27me3 ChIP-seq signal for each of the tier 1 and 2 cell types. For DBX1, representative of a set of primarily neuronal transcription factors associated with unit 26,31, there is a repressive H3K27me3 signal in both H1 hESCs and HUVECs; for IRX6, representative of a set of body patterning transcription factors associated with SOM unit 27,30, the repressive mark is restricted largely to the embryonic stem (ES) cell. An interactive version of this figure is available in the online version of the paper.
chromatin data. Many of the ultra-fine-grained state classifications revealed in the SOM are associated with specific gene ontology (GO) terms (right panel of Fig. 7c). For instance, the left panel of Fig. 7c identifies ten SOM map units enriched with genomic regions associated with genes associated with the GO term immune response. The central panel identifies a different set of map units enriched for the GO term sequence-specific transcription factor activity. The two map units most enriched for this GO term, indicated by the darkest green colouring, contain genes with segments that are high in
Genes above threshold
Genes above threshold
H3K27me3 in H1 hESCs, but that differ in H3K27me3 levels in HUVECs. Gene function analysis with the GO ontology tool (GREAT71) reveals that the map unit with high H3K27me3 levels in both cell types is enriched in transcription factor genes with known neuronal functions, whereas the neighbouring map unit is enriched in genes involved in body patterning. The genome browser shots at the bottom of Fig. 7c pick out an example region for each of the two SOM map units illustrating the difference in H3K27me3 signal. Overall, we have 228 distinct GO terms associated with specific segments across
RESEARCH ARTICLE
one or more states (A. Mortazavi, personal communication), and can assign over one-third of genes to a GO annotation solely on the basis of its multicellular histone patterns. Thus, the SOM analysis provides a fine-grained map of chromatin data across multiple cell types, which can then be used to relate chromatin structure to other data types at differing levels of resolution (for instance, the large cluster of units containing any active TSS, its subclusters composed of units enriched in TSSs active in only one cell type, or individual map units significantly enriched for specific GO terms). The classifications presented here are necessarily limited by the assays and cell lines studied, and probably contain a number of heterogeneous classes of elements. Nonetheless, robust classifications can be made, allowing a systematic view of the human genome.
a
138960000
NACC2 Transcription All reads
138970000
138980000
138990000
POLR2A
Paternal only
0
Maternal only
5
All reads
H3K79me2
Paternal only
0
Maternal only
5
All reads
H3K27me3
Insights into human genomic variation

We next explored the potential impact of sequence variation on ENCODE functional elements. We examined allele-specific variation using results from the GM12878 cells that are derived from an individual (NA12878) sequenced in the 1000 Genomes project, along with her parents. Because ENCODE assays are predominantly sequencebased, the trio design allows each GM12878 data set to be divided by the specific parental contributions at heterozygous sites, producing aggregate haplotypic signals from multiple genomic sites. We examined 193 ENCODE assays for allele-specific biases using 1,409,992 phased, heterozygous SNPs and 167,096 insertions/deletions (indels) (Fig. 8). Alignment biases towards alleles present in the reference genome sequence were avoided using a sequence specifically tailored to the variants and haplotypes present in NA12878 (a personalized genome)72. We found instances of preferential binding towards each parental allele. For example, comparison of the results from the POLR2A, H3K79me2 and H3K27me3 assays in the region of NACC2 (Fig. 8a) shows a strong paternal bias for H3K79me2 and POL2RA and a strong maternal bias for H3K27me3, indicating differential activity for the maternal and paternal alleles. Figure 8b shows the correlation of selected allele-specific signals across the whole genome. For instance, we found a strong allelic correlation between POL2RA and BCLAF1 binding, as well as negative correlation between H3K79me2 and H3K27me3, both at genes (Fig. 8b, below the diagonal, bottom left) and chromosomal segments (top right). Overall, we found that positive allelic correlations among the 193 ENCODE assays are stronger and more frequent than negative correlations. This may be due to preferential capture of accessible alleles and/or the specific histone modification and transcription factor, assays used in the project.
Paternal only
0
Maternal only
5
Common SNPs
H3K4me1 H3K4me2 H3K4me3 H2AFZ H3K9ac H3K27ac H3K79me2 H3K36me3 H3K9me3 H3K27me3 DNase POLR2A POLR2A elongating CTCF BATF BCLAF1 ETS1 MEF2A MTA3 EP300 POU2F2 RAD21 r2 RFX5 0.383 RUNX3 SMC3 STAT5A 0 TAF1 TBP YY1 0.020 ZNF143
Rare variants, individual genomes and somatic variants

We further investigated the potential functional effects of individual variation in the context of ENCODE annotations. We divided NA12878 variants into common and rare classes, and partitioned these into those overlapping ENCODE annotation (Fig. 9a and Supplementary Tables 1 and 2, section K). We also predicted potential functional effects: for protein-coding genes, these are either nonsynonymous SNPs or variants likely to induce loss of function by frame-shift, premature stop, or splice-site disruption; for other regions, these are variants that overlap a transcription-factorbinding site. We found similar numbers of potentially functional variants affecting protein-coding genes or affecting other ENCODE annotations, indicating that many functional variants within individual genomes lie outside exons of protein-coding genes. A more detailed analysis of regulatory variant annotation is described in ref. 73. To study further the potential effects of NA12878 genome variants on transcription-factor-binding regions, we performed peak calling using a constructed personal diploid genome sequence for NA12878 (ref. 72). We aligned ChIP-seq sequences from GM12878 separately against the maternal and paternal haplotypes. As expected, a greater
Figure 8 | Allele-specific ENCODE elements. a, Representative allele-specific information from GM12878 cells for selected assays around the first exon of the NACC2 gene (genomic region Chr9: 138950000138995000, GRCh37). Transcription signal is shown in green, and the three sections show allelespecific data for three data sets (POLR2A, H3K79me2 and H3K27me3 ChIPseq). In each case the purple signal is the processed signal for all sequence reads for the assay, whereas the blue and red signals show sequence reads specifically assigned to either the paternal or maternal copies of the genome, respectively. The set of common SNPs from dbSNP, including the phased, heterozygous SNPs used to provide the assignment, are shown at the bottom of the panel. NACC2 has a statistically significant paternal bias for POLR2A and the transcription-associated mark H3K79me2, and has a significant maternal bias for the repressive mark H3K27me3. b, Pair-wise correlations of allele-specific signal within single genes (below the diagonal) or within individual ChromHMM segments across the whole genome for selected DNase-seq and histone modification and transcription factor ChIP-seq assays. The extent of correlation is coloured according to the heat-map scale indicated from positive correlation (red) through to anti-correlation (blue). An interactive version of this figure is available in the online version of the paper.
fraction of reads were aligned than to the reference genome (see Supplementary Information, Supplementary Fig. 1, section K). On average, approximately 1% of transcription-factor-binding sites in GM12878 cells are detected in a haplotype-specific fashion. For instance, Fig. 9b shows a CTCF-binding site not detected using the
H3K4me1 H3K4me2 H3K4me3 H2AFZ H3K9ac H3K27ac H3K79me2 H3K36me3 H3K9me3 H3K27me3 DNase POLR2A POLR2A elongating CTCF BATF BCLAF1 ETS1 MEF2A MTA3 EP300 POU2F2 RAD21 RFX5 RUNX3 SMC3 STAT5A TAF1 TBP YY1 ZNF143
ARTICLE RESEARCH
a
All variants in NA12878 (2,998,908)
Protein-coding annotation with predicted functional effect ENCODE non-coding annotation with predicted functional effect
Annotated (473,356)
Rare (86,420)
Rare annotated (23,227)
Rare variants with predicted functional effect (1,482)
b
20 15
Common annotated (450,129)
Common variants with predicted functional effect (27,940)
c
Variants per megabase of DHS relative to null Paternal Maternal
20 10 0
Non-genic TSS-distal cell-specific DHS peaks
Read count
10
P < 0.05
Depleted Enriched NA
10 20
5
126665000 126666000 126667000 126668000 126669000 126670000
10
Genome position on chromosome 10
Figure 9 | Examining ENCODE elements on a per individual basis in the normal and cancer genome. a, Breakdown of variants in a single genome (NA12878) by both frequency (common or rare (that is, variants not present in the low-coverage sequencing of 179 individuals in the pilot 1 European panel of the 1000 Genomes project55)) and by ENCODE annotation, including proteincoding gene and non-coding elements (GENCODE annotations for proteincoding genes, pseudogenes and other ncRNAs, as well as transcription-factorbinding sites from ChIP-seq data sets, excluding broad annotations such as histone modifications, segmentations and RNA-seq). Annotation status is further subdivided by predicted functional effect, being non-synonymous and missense mutations for protein-coding regions and variants overlapping bound
reference sequence that is only present on the paternal haplotype due to a 1-bp deletion (see also Supplementary Fig. 2, section K). As costs of DNA sequencing decrease further, optimized analysis of ENCODE-type data should use the genome sequence of the individual or cell being analysed when possible. Most analyses of cancer genomes so far have focused on characterizing somatic variants in protein-coding regions. We intersected four available whole-genome cancer data sets with ENCODE annotations (Fig. 9c and Supplementary Fig. 2, section L). Overall, somatic variation is relatively depleted from ENCODE annotated regions, particularly for elements specific to a cell type matching the putative tumour source (for example, skin melanocytes for melanoma). Examining the mutational spectrum of elements in introns for cases where a strand-specific mutation assignment could be made reveals that there are mutational spectrum differences between DHSs and unannotated regions (0.06 Fishers exact test, Supplementary Fig. 3, section L). The suppression of somatic mutation is consistent with important functional roles of these elements within tumour cells, highlighting a potential alternative set of targets for examination in cancer.
Common variants associated with disease

In recent years, GWAS have greatly extended our knowledge of genetic loci associated with human disease risk and other phenotypes.
H S H MM U M VE e St lan C e o U Hllate bi 9 qu E Pr ito S o us H gFi ep b Ao G SM H i 2 C PDE HTPS se 6 R ru E6 8 m E M fr 7 e e O du e st llo e p ob Fi HT l br E 89 obl Pa n- 88 . is t l M HM ets yo EC H ep LNme at C tr oc aP y N te C H s G ho EK M ri 12 on H 8 1H 7 ES 8 H MC C eL F U a-S7 RO 3 t K5 sa A562 49 C ll
126671000
DHS cell type
transcription factor motifs for non-coding element annotations. A substantial proportion of variants are annotated as having predicted functional effects in the non-coding category. b, One of several relatively rare occurrences, where alignment to an individual genome sequence (paternal and maternal panels) shows a different readout from the reference genome. In this case, a paternalhaplotype-specific CTCF peak is identified. c, Relative level of somatic variants from a whole-genome melanoma sample that occur in DHSs unique to different cell lines. The coloured bars show cases that are significantly enriched or suppressed in somatic mutations. Details of ENCODE cell types can be found at http://encodeproject.org/ENCODE/cellTypes.html. An interactive version of this figure is available in the online version of the paper.
The output of these studies is a series of SNPs (GWAS SNPs) correlated with a phenotype, although not necessarily the functional variants. Notably, 88% of associated SNPs are either intronic or intergenic74. We examined 4,860 SNPphenotype associations for 4,492 SNPs curated in the National Human Genome Research Institute (NHGRI) GWAS catalogue74. We found that 12% of these SNPs overlap transcription-factor-occupied regions whereas 34% overlap DHSs (Fig. 10a). Both figures reflect significant enrichments relative to the overall proportions of 1000 Genomes project SNPs (about 6% and 23%, respectively). Even after accounting for biases introduced by selection of SNPs for the standard genotyping arrays, GWAS SNPs show consistently higher overlap with ENCODE annotations (Fig. 10a, see Supplementary Information). Furthermore, after partitioning the genome by density of different classes of functional elements, GWAS SNPs were consistently enriched beyond all the genotyping SNPs in function-rich partitions, and depleted in function-poor partitions (see Supplementary Fig. 1, section M). GWAS SNPs are particularly enriched in the segmentation classes associated with enhancers and TSSs across several cell types (see Supplementary Fig. 2, section M). Examining the SOM of integrated ENCODE annotations (see above), we found 19 SOM map units showing significant enrichment for GWAS SNPs, including many SOM units previously associated with specific gene functions, such as the immune response regions.
RESEARCH ARTICLE
Thus, an appreciable proportion of SNPs identified in initial GWAS scans are either functional or lie within the length of an ENCODE annotation (,500 bp on average) and represent plausible candidates for the functional variant. Expanding the set of feasible functional SNPs to those in reasonable linkage disequilibrium, up to 71% of GWAS SNPs have a potential causative SNP overlapping a DNase I site, and 31% of loci have a candidate SNP that overlaps a binding site occupied by a transcription factor (see also refs 73, 75). The GWAS catalogue provides a rich functional categorization from the precise phenotypes being studied. These phenotypic categorizations are nonrandomly associated with ENCODE annotations and there is marked correspondence between the phenotype and the identity of the cell type or transcription factor used in the ENCODE assay (Fig. 10b). For example, five SNPs associated with Crohns disease overlap GATA2-binding sites (P value 0.003 by random permutation or 0.001 by an empirical approach comparing to the GWAS-matched SNPs; see Supplementary Information), and fourteen are located in DHSs found in immunologically relevant cell
a
Fraction of SNPs that overlap feature
types. A notable example is a gene desert on chromosome 5p13.1 containing eight SNPs associated with inflammatory diseases. Several are close to or within DHSs in T-helper type 1 (TH1) and TH2 cells as well as peaks of binding by transcription factors in HUVECs (Fig. 10c). The latter cell line is not immunological, but factor occupancy detected there could be a proxy for binding of a more relevant factor, such as GATA3, in T cells. Genetic variants in this region also affect expression levels of PTGER4 (ref. 76), encoding the prostaglandin receptor EP4. Thus, the ENCODE data reinforce the hypothesis that genetic variants in 5p13.1 modulate the expression of flanking genes, and furthermore provide the specific hypothesis that the variants affect occupancy of a GATA factor in an allelespecific manner, thereby influencing susceptibility to Crohns disease. Nonrandom association of phenotypes with ENCODE cell types strengthens the argument that at least some of the GWAS lead SNPs are functional or extremely close to functional variants. Each of the associations between a lead SNP and an ENCODE annotation remains a credible hypothesis of a particular functional element
Overlap any TF occupancy
SNP-pheno associations
GM12878 POLR2A
GM12878 MEF2A
GM12878 NFKB1
HeLa-S3 CEBPB
GM12878 EGR1
GM12878 EBF1
GM12878 BATF
GM12878 PAX5
GM12878 SPI1
GM12878 IRF4
HUVEC GATA2
Phenotype
0.2
0.1
DNase I peaks Transcription factor
Total 4,860 600 Multiple sclerosis 71 15 Systemic lupus erythematosus 62 10 Height 204 34 Rheumatoid arthritis 57 11 Chronic lymphocytic leukaemia 17 8 Celiac disease 54 11 Ulcerative colitis 85 11 Crohns disease 105 20
47 4 4 3 3 1 1 3 2
78 4 4 7 4 1 4 2 2
57 3 6 3 2 4 3 3 2
35 2 4 2 4 1 2 3 1
35 3 3 5 0 1 1 1 1
54 4 1 3 4 2 1 3 2
47 3 1 1 4 3 2 2 2
45 2 4 5 2 1 1 2 1
29 2 1 0 0 0 0 2 5
28 1 2 6 1 0 0 2 1
69 1 2 7 1 0 0 0 2
54 0 1 6 0 0 0 1 2
85 192 57 1 5 4 2 4 2 6 9 3 2 11 3 0 2 0 0 2 1 2 7 2 6 9 5
Chr5:
39500000
Human Feb. 2009 (GRCh37/hg19) Chr5: 3927450140819500 (1,545,000 bp) 40000000 40500000
C9 DAB2 BC026261 GWAS catalogue
PTGER4 TTC33 OSRF PRKAA1
Chr5: 4039000140440000 (50,000 bp)
Crohns disease Ulcerative colitis Multiple sclerosis HUVEC GATA2 Transcription factors HUVEC input HUVEC DNase I TH1 TH2
rs4613763
rs17234657
rs11742570 rs1992660 rs6451493
rs6896969
rs1373692
rs9292777
Figure 10 | Comparison of genome-wide-association-study-identified loci with ENCODE data. a, Overlap of lead SNPs in the NHGRI GWAS SNP catalogue (June 2011) with DHSs (left) or transcription-factor-binding sites (right) as red bars compared with various control SNP sets in blue. The control SNP sets are (from left to right): SNPs on the Illumina 2.5M chip as an example of a widely used GWAS SNP typing panel; SNPs from the 1000 Genomes project; SNPs extracted from 24 personal genomes (see personal genome variants track at http://main.genome-browser.bx.psu.edu (ref. 80)), all shown as blue bars. In addition, a further control used 1,000 randomizations from the genotyping SNP panel, matching the SNPs with each NHGRI catalogue SNP for allele frequency and distance to the nearest TSS (light blue bars with bounds at 1.5 times the interquartile range). For both DHSs and transcription-factorbinding regions, a larger proportion of overlaps with GWAS-implicated SNPs is found compared to any of the controls sets. b, Aggregate overlap of
phenotypes to selected transcription-factor-binding sites (left matrix) or DHSs in selected cell lines (right matrix), with a count of overlaps between the phenotype and the cell line/factor. Values in blue squares pass an empirical P-value threshold #0.01 (based on the same analysis of overlaps between randomly chosen, GWAS-matched SNPs and these epigenetic features) and have at least a count of three overlaps. The P value for the total number of phenotypetranscription factor associations is ,0.001. c, Several SNPs associated with Crohns disease and other inflammatory diseases that reside in a large gene desert on chromosome 5, along with some epigenetic features indicative of function. The SNP (rs11742570) strongly associated to Crohns disease overlaps a GATA2 transcription-factor-binding signal determined in HUVECs. This region is also DNase I hypersensitive in HUVECs and T-helper TH1 and TH2 cells. An interactive version of this figure is available in the online version of the paper.
hTH1 all
0.3
hTH2 DS7842
HepG2 CTCF
HUVEC all
ARTICLE RESEARCH
class or cell type to explore with future experiments. Supplementary Tables 13, section M, list all 14,885 pairwise associations across the ENCODE annotations. The accompanying papers have a more detailed examination of common variants with other regulatory information19,25,29,73,75,77. So far, ENCODE has sampled 119 of 1,800 known transcription factors and general components of the transcriptional machinery on a limited number of cell types, and 13 of more than 60 currently known histone or DNA modifications across 147 cell types. DNase I, FAIRE and extensive RNA assays across subcellular fractionations have been undertaken on many cell types, but overall these data reflect a minor fraction of the potential functional information encoded in the human genome. An important future goal will be to enlarge this data set to additional factors, modifications and cell types, complementing the other related projects in this area (for example, Roadmap Epigenomics Project, http:// www.roadmapepigenomics.org/, and International Human Epigenome Consortium, http://www.ihec-epigenomes.org/). These projects will constitute foundational resources for human genomics, allowing a deeper interpretation of the organization of gene and regulatory information and the mechanisms of regulation, and thereby provide important insights into human health and disease. Co-published ENCODE-related papers can be explored online via the Nature ENCODE explorer (http://www.nature.com/ENCODE), a specially designed visualization tool that allows users to access the linked papers and investigate topics that are discussed in multiple papers via thematically organized threads.
Concluding remarks
The unprecedented number of functional elements identified in this study provides a valuable resource to the scientific community as well as significantly enhances our understanding of the human genome. Our analyses have revealed many novel aspects of gene expression and regulation as well as the organization of such information, as illustrated by the accompanying papers (see http://www.encodeproject. org/ENCODE/pubs.html for collected ENCODE publications). However, there are still many specific details, particularly about the mechanistic processes that generate these elements and how and where they function, that require additional experiments to elucidate. The large spread of coveragefrom our highest resolution, most conservative set of bases implicated in GENCODE protein-coding gene exons (2.9%) or specific protein DNA binding (8.5%) to the broadest, most general set of marks covering the genome (approximately 80%), with many gradations in betweenpresents a spectrum of elements with different functional properties discovered by ENCODE. A total of 99% of the known bases in the genome are within 1.7 kb of any ENCODE element, whereas 95% of bases are within 8 kb of a bound transcription factor motif or DNase I footprint. Interestingly, even using the most conservative estimates, the fraction of bases likely to be involved in direct gene regulation, even though incomplete, is significantly higher than that ascribed to proteincoding exons (1.2%), raising the possibility that more information in the human genome may be important for gene regulation than for biochemical function. Many of the regulatory elements are not constrained across mammalian evolution, which so far has been one of the most reliable indications of an important biochemical event for the organism. Thus, our data provide orthologous indicators for suggesting possible functional elements. Importantly, for the first time we have sufficient statistical power to assess the impact of negative selection on primate-specific elements, and all ENCODE classes display evidence of negative selection in these unique-to-primate elements. Furthermore, even with our most conservative estimate of functional elements (8.5% of putative DNA/protein binding regions) and assuming that we have already sampled half of the elements from our transcription factor and cell-type diversity, one would estimate that at a minimum 20% (17% from protein binding and 2.9% protein coding gene exons) of the genome participates in these specific functions, with the likely figure significantly higher. The broad coverage of ENCODE annotations enhances our understanding of common diseases with a genetic component, rare genetic diseases, and cancer, as shown by our ability to link otherwise anonymous associations to a functional element. ENCODE and similar studies provide a first step towards interpreting the rest of the genomebeyond protein-coding genesthereby augmenting common disease genetic studies with testable hypotheses. Such information justifies performing whole-genome sequencing (rather than exome only, 1.2% of the genome) on rare diseases and investigating somatic variants in non-coding functional elements, for instance, in cancer. Furthermore, as GWAS analyses typically associate disease to SNPs in large regions, comparison to ENCODE noncoding functional elements can help pinpoint putative causal variants in addition to refinement of location by fine-mapping techniques78. Combining ENCODE data with allele-specific information derived from individual genome sequences provides specific insight on the impact of a genetic variant. Indeed, we believe that a significant goal would be to use functional data such as that derived from this project to assign every genomic variant to its possible impact on human phenotypes.
METHODS SUMMARY
For full details of Methods, see Supplementary Information.
Received 24 November 2011; accepted 29 May 2012. 1. 2. 3. 4. 5. ENCODE Project Consortium. The ENCODE (ENCyclopedia Of DNA Elements) Project. Science 306, 636640 (2004). Birney, E. et al. Identification and analysis of functional elements in 1% of the human genome by the ENCODE pilot project. Nature 447, 799816 (2007). The ENCODE Project Consortium. A users guide to the encyclopedia of DNA elements (ENCODE). PLoS Biol. 9, e1001046 (2011). Mouse Genome Sequencing Consortium. Initial sequencing and comparative analysis of the mouse genome. Nature 420, 520562 (2002). Chiaromonte, F. et al. The share of human genomic DNA under selection estimated from human-mouse genomic alignments. Cold Spring Harb. Symp. Quant. Biol. 68, 245254 (2003). Cooper, G. M. et al. Distribution and intensity of constraint in mammalian genomic sequence. Genome Res. 15, 901913 (2005). Parker, S. C., Hansen, L., Abaan, H. O., Tullius, T. D. & Margulies, E. H. Local DNA topography correlates with functional noncoding regions of the human genome. Science 324, 389392 (2009). Lindblad-Toh, K. et al. A high-resolution map of human evolutionary constraint using 29 mammals. Nature 478, 476482 (2011). Pheasant, M. & Mattick, J. S. Raising the estimate of functional human sequences. Genome Res. 17, 12451253 (2007). Ponting, C. P. & Hardison, R. C. What fraction of the human genome is functional? Genome Res. 21, 17691776 (2011). Asthana, S. et al. Widely distributed noncoding purifying selection in the human genome. Proc. Natl Acad. Sci. USA 104, 1241012415 (2007). Landt, S. G. et al. ChIP-seq guidelines and practices used by the ENCODE and modENCODE consortia. Genome Res. http://dx.doi.org/10.1101/gr.136184.111 (2012). Li, Q., Brown, J. B., Huang, H. & Bickel, P. J. Measuring reproducibility of highthroughput experiments. Ann. Appl. Stat. 5, 17521779 (2011). Harrow, J. et al. GENCODE: The reference human genome annotation for the ENCODE project. Genome Res. http://dx.doi.org/10.1101/gr.135350.111 (2012). Howald, C. et al. Combining RT-PCR-seq and RNA-seq to catalog all genic elements encoded in the human genome. Genome Res. http://dx.doi.org/10.1101/ gr.134478.111 (2012). Djebali, S. et al. Landscape of transcription in human cells. Nature http:// dx.doi.org/10.1038/nature11233 (this issue). Derrien, T. et al. The GENCODE v7 catalog of human long noncoding RNAs: Analysis of their gene structure, evolution, and expression. Genome Res. http:// dx.doi.org/10.1101/gr.132159.111 (2012). Pei, B. et al. The GENCODE pseudogene resource. Genome Biol. 13, R51 (2012). Gerstein, M. B. et al. Architecture of the human regulatory network derived from ENCODE data. Nature http://dx.doi.org/10.1038/nature11245 (this issue). Bickel, P. J., Boley, N., Brown, J. B., Huang, H. Y. & Zhang, N. R. Subsampling methods for genomic inference. Ann. Appl. Stat. 4, 16601697 (2010). Kaplan, T. et al. Quantitative models of the mechanisms that control genome-wide patterns of transcription factor binding during early Drosophila development. PLoS Genet. 7, e1001290 (2011). Li, X. Y. et al. The role of chromatin accessibility in directing the widespread, overlapping patterns of Drosophila transcription factor binding. Genome Biol. 12, R34 (2011).
6. 7.
8. 9. 10. 11. 12.
13. 14.
15.
16. 17.
18. 19. 20. 21.
22.
RESEARCH ARTICLE
23. Pique-Regi, R. et al. Accurate inference of transcription factor binding from DNA sequence and chromatin accessibility data. Genome Res. 21, 447455 (2011). 24. Zhang, Y. et al. Primary sequence and epigenetic determinants of in vivo occupancy of genomic DNA by GATA1. Nucleic Acids Res. 37, 70247038 (2009). 25. Neph, S. et al. An expansive human regulatory lexicon encoded in transcription factor footprints. Nature http://dx.doi.org/10.1038/nature11212 (this issue). 26. Whitfield, T. W. et al. Functional analysis of transcription factor binding sites in human promoters. Genome Biol. 13, R50 (2012). 27. Gross, D. S. & Garrard, W. T. Nuclease hypersensitive sites in chromatin. Annu. Rev. Biochem. 57, 159197 (1988). 28. Urnov, F. D. Chromatin remodeling as a guide to transcriptional regulatory networks in mammals. J. Cell. Biochem. 88, 684694 (2003). 29. Thurman, R. E. et al. The accessible chromatin landscape of the human genome. Nature http://dx.doi.org/10.1038/nature11232 (this issue). 30. Kundaje, A. et al. Ubiquitous heterogeneity and asymmetry of the chromatin environment at regulatory elements. Genome Res. http://dx.doi.org/10.1101/ gr.136366.111 (2012). 31. Schultz, D. C., Ayyanathan, K., Negorev, D., Maul, G. G. & Rauscher, F. J. III. SETDB1: a novel KAP-1-associated histone H3, lysine 9-specific methyltransferase that contributes to HP1-mediated silencing of euchromatic genes by KRAB zinc-finger proteins. Genes Dev. 16, 919932 (2002). 32. Frietze, S., OGeen, H., Blahnik, K. R., Jin, V. X. & Farnham, P. J. ZNF274 recruits the histone methyltransferase SETDB1 to the 39 ends of ZNF genes. PLoS ONE 5, e15082 (2010). 33. Boyle, A. P. et al. High-resolution genome-wide in vivo footprinting of diverse transcription factors in human cells. Genome Res. 21, 456464 (2011). 34. Hesselberth, J. R. et al. Global mapping of protein-DNA interactions in vivo by digital genomic footprinting. Nature Methods 6, 283289 (2009). 35. Zhang, Y. et al. Model-based analysis of ChIP-Seq (MACS). Genome Biol. 9, R137 (2008). 36. Kouzarides, T. Chromatin modifications and their function. Cell 128, 693705 (2007). 37. Li, B., Carey, M. & Workman, J. L. The role of chromatin during transcription. Cell 128, 707719 (2007). 38. Hon, G. C., Hawkins, R. D. & Ren, B. Predictive chromatin signatures in the mammalian genome. Hum. Mol. Genet. 18, R195R201 (2009). 39. Zhou, V. W., Goren, A. & Bernstein, B. E. Charting histone modifications and the functional organization of mammalian genomes. Nature Rev. Genet. 12, 718 (2011). 40. Ernst, J. et al. Mapping and analysis of chromatin state dynamics in nine human cell types. Nature 473, 4349 (2011). 41. Hon, G., Wang, W. & Ren, B. Discovery and annotation of functional chromatin signatures in the human genome. PLoS Comput. Biol. 5, e1000566 (2009). 42. Ball, M. P. et al. Targeted and genome-scale strategies reveal gene-body methylation signatures in human cells. Nature Biotechnol. 27, 361368 (2009). 43. Meissner, A. et al. Genome-scale DNA methylation maps of pluripotent and differentiated cells. Nature 454, 766770 (2008). 44. Ogryzko, V. V., Schiltz, R. L., Russanova, V., Howard, B. H. & Nakatani, Y. The transcriptional coactivators p300 and CBP are histone acetyltransferases. Cell 87, 953959 (1996). 45. Lister, R. et al. Human DNA methylomes at base resolution show widespread epigenomic differences. Nature 462, 315322 (2009). 46. Dekker, J. Gene regulation in the third dimension. Science 319, 17931794 (2008). 47. Dostie, J. et al. Chromosome Conformation Capture Carbon Copy (5C): a massively parallel solution for mapping interactions between genomic elements. Genome Res. 16, 12991309 (2006). 48. Lajoie, B. R., van Berkum, N. L., Sanyal, A. & Dekker, J. My5C: web tools for chromosome conformation capture studies. Nature Methods 6, 690691 (2009). 49. Sanyal, A., Lajoie, B., Jain, G. & Dekker, J. The long-range interaction landscape of gene promoters. Nature http://dx.doi.org/10.1038/nature11279 (this issue). 50. Fullwood, M. J. et al. An oestrogen-receptor-alpha-bound human chromatin interactome. Nature 462, 5864 (2009). 51. Li, G. et al. Extensive promoter-centered chromatin interactions provide a topological basis for transcription regulation. Cell 148, 8498 (2012). 52. Borneman, A. R. et al. Divergence of transcription factor binding sites across related yeast species. Science 317, 815819 (2007). 53. Odom, D. T. et al. Tissue-specific transcriptional regulation has diverged significantly between human and mouse. Nature Genet. 39, 730732 (2007). 54. Schmidt, D. et al. Five-vertebrate ChIP-seq reveals the evolutionary dynamics of transcription factor binding. Science 328, 10361040 (2010). 55. A map of human genome variation from population-scale sequencing. Nature 467, 10611073 (2010). 56. King, M. C. & Wilson, A. C. Evolution at two levels in humans and chimpanzees. Science 188, 107116 (1975). 57. Spivakov, M. et al. Analysis of variation at transcription factor binding sites in Drosophila and humans. Genome Biol. 13, R49 (2012). 58. Sandelin, A. et al. Mammalian RNA polymerase II core promoters: insights from genome-wide studies. Nature Rev. Genet. 8, 424436 (2007). 59. Dong, X. et al. Modeling gene expression using chromatin features in various cellular contexts. Genome Biol. 13, R53 (2012). 60. Huff, J. T., Plocik, A. M., Guthrie, C. & Yamamoto, K. R. Reciprocal intronic and exonic histone modification regions in humans. Nature Struct. Mol. Biol. 17, 14951499 (2010). 61. Tilgner, H. et al. Deep sequencing of subcellular RNA fractions shows splicing to be predominantly co-transcriptional in the human genome but inefficient for lncRNAs. Genome Res. http://dx.doi.org/10.1101/gr.134445.111 (2012).
62. Fu, Y., Sinha, M., Peterson, C. L. & Weng, Z. The insulator binding protein CTCF positions 20 nucleosomes around its binding sites across the human genome. PLoS Genet. 4, e1000138 (2008). 63. Kornberg, R. D. & Stryer, L. Statistical distributions of nucleosomes: nonrandom locations by a stochastic mechanism. Nucleic Acids Res. 16, 66776690 (1988). 64. Schones, D. E. et al. Dynamic regulation of nucleosome positioning in the human genome. Cell 132, 887898 (2008). 65. Valouev, A. et al. Determinants of nucleosome organization in primary human cells. Nature 474, 516520 (2011). 66. Frietze, S. et al. Cell type-specific binding patterns reveal that TCF7L2 can be tethered to the genome by association with GATA3. Genome Biol. 13, R52 (2012). 67. Yip, K. Y. et al. Classification of human genomic regions based on experimentallydetermined binding sites of more than 100 transcription-related factors. Genome Biol. 13, R48 (2012). 68. Hoffman, M. M. et al. Unsupervised pattern discovery in human chromatin structure through genomic segmentation. Nature Methods 9, 473476 (2012). 69. Kapranov, P. et al. RNA maps reveal new RNA classes and a possible function for pervasive transcription. Science 316, 14841488 (2007). 70. Koch, F. et al. Transcription initiation platforms and GTF recruitment at tissuespecific enhancers and promoters. Nature Struct. Mol. Biol. 18, 956963 (2011). 71. McLean, C. Y. et al. GREAT improves functional interpretation of cis-regulatory regions. Nature Biotechnol. 28, 495501 (2010). 72. Rozowsky, J. et al. AlleleSeq: analysis of allele-specific expression and binding in a network framework. Mol. Syst. Biol. 7, 522 (2011). 73. Boyle, A. P. et al. Annotation of functional variation in personal genomes using RegulomeDB. Genome Res. http://dx.doi.org/10.1101/gr.137323.112 (2012). 74. Hindorff, L. A. et al. Potential etiologic and functional implications of genome-wide association loci for human diseases and traits. Proc. Natl Acad. Sci. USA 106, 93629367 (2009). 75. Schaub, M. A., Boyle, A. P., Kundaje, A., Batzoglou, S. & Snyder, M. Linking disease associations with regulatory information in the human genome. Genome Res. http://dx.doi.org/10.1101/gr.136127.111 (2012). 76. Libioulle, C. et al. Novel Crohn disease locus identified by genome-wide association maps to a gene desert on 5p13.1 and modulates expression of PTGER4. PLoS Genet. 3, e58 (2007). 77. Vernot, B. et al. Personal and population genomics of human regulatory variation. Genome Res. http://dx.doi.org/10.1101/gr.134890.111 (2012). 78. Harismendy, O. et al. 9p21 DNA variants associated with coronary artery disease impair interferon-c signalling response. Nature 470, 264268 (2011). 79. Cheng, C. et al. Understanding transcriptional regulation by integrative analysis of transcription factor binding data. Genome Res. http://dx.doi.org/10.1101/ gr.136838.111 (2012). 80. Schuster, S. C. et al. Complete Khoisan and Bantu genomes from southern Africa. Nature 463, 943947 (2010). Supplementary Information is available in the online version of the paper. Acknowledgements We thank additional members of our laboratories and institutions who have contributed to the experimental and analytical components of this project. We thank D. Leja for assistance with production of the figures. The Consortium is funded by grants from the NHGRI as follows: production grants: U54HG004570 (B. E. Bernstein); U01HG004695 (E. Birney); U54HG004563 (G. E. Crawford); U54HG004557 (T. R. Gingeras); U54HG004555 (T. J. Hubbard); U41HG004568 (W. J. Kent); U54HG004576 (R. M. Myers); U54HG004558 (M. Snyder); U54HG004592 (J. A. Stamatoyannopoulos). Pilot grants: R01HG003143 (J. Dekker); RC2HG005591 and R01HG003700 (M. C. Giddings); R01HG004456-03 (Y. Ruan); U01HG004571 (S. A. Tenenbaum); U01HG004561 (Z. Weng); RC2HG005679 (K. P. White). This project was supported in part by American Recovery and Reinvestment Act (ARRA) funds from the NHGRI through grants U54HG004570, U54HG004563, U41HG004568, U54HG004592, R01HG003143, RC2HG005591, R01HG003541, U01HG004561, RC2HG005679 and R01HG003988 (L. Pennacchio). In addition, work from NHGRI Groups was supported by the Intramural Research Program of the NHGRI (L. Elnitski, ZIAHG200323; E. H. Margulies, ZIAHG200341). Research in the Pennachio laboratory was performed at Lawrence Berkeley National Laboratory and at the United States Department of Energy Joint Genome Institute, Department of Energy Contract DE-AC02-05CH11231, University of California. Author Contributions See the consortium author list for details of author contributions. Author Information The Supplementary Information is accompanied by a Virtual Machine (VM) containing the functioning analysis data and code. Further details of the VM are available from http://encodeproject.org/ENCODE/integrativeAnalysis/VM. Reprints and permissions information is available at www.nature.com/reprints. This paper is distributed under the terms of the Creative Commons Attribution-Non-Commercial-Share Alike licence, and the online version of the paper is freely available to all readers. The authors declare no competing financial interests. Readers are welcome to comment on the online version of the paper. Correspondence and requests for materials should be addressed to E.B. ([email protected]).
The ENCODE Project Consortium Overall coordination (data analysis coordination) Ian Dunham1, Anshul Kundaje2{; Data production leads (data production) Shelley F. Aldred3, Patrick J. Collins3, Carrie A. Davis4, Francis Doyle5, Charles B. Epstein6, Seth Frietze7, Jennifer Harrow8, Rajinder Kaul9, Jainab Khatun10, Bryan R. Lajoie11, Stephen G. Landt12, Bum-Kyu Lee13,
ARTICLE RESEARCH
Florencia Pauli14, Kate R. Rosenbloom15, Peter Sabo16, Alexias Safi17, Amartya Sanyal11, Noam Shoresh6, Jeremy M. Simon18, Lingyun Song17, Nathan D. Trinklein3; Lead analysts (data analysis) Robert C. Altshuler19, Ewan Birney1, James B. Brown20, Chao Cheng21, Sarah Djebali22, Xianjun Dong23, Ian Dunham1, Jason Ernst19{, Terrence S. Furey24, Mark Gerstein21, Belinda Giardine25, Melissa Greven23, Ross C. Hardison25,26, Robert S. Harris25, Javier Herrero1, Michael M. Hoffman16, Sowmya Iyer27, Manolis Kellis19, Jainab Khatun10, Pouya Kheradpour19, Anshul Kundaje2{, Timo Lassmann28, Qunhua Li20{, Xinying Lin23, Georgi K. Marinov29, Angelika Merkel22, Ali Mortazavi30, Stephen C. J. Parker31, Timothy E. Reddy14{, Joel Rozowsky21, Felix Schlesinger4, Robert E. Thurman16, Jie Wang23, Lucas D. Ward19, Troy W. Whitfield23, Steven P. Wilder1, Weisheng Wu25, Hualin S. Xi32, Kevin Y. Yip21{, Jiali Zhuang23; Writing group Bradley E. Bernstein6,33, Ewan Birney1, Ian Dunham1, Eric D. Green34, Chris Gunter14, Michael Snyder12; NHGRI project management (scientific management) Michael J. Pazin35, Rebecca F. Lowdon35{, Laura A. L. Dillon35{, Leslie B. Adams35, Caroline J. Kelly35, Julia Zhang35{, Judith R. Wexler35{, Eric D. Green34, Peter J. Good35, Elise A. Feingold35; Principal investigators (steering committee) Bradley E. Bernstein6,33, Ewan Birney1, Gregory E. Crawford17,36, Job Dekker11, Laura Elnitski37, Peggy J. Farnham7, Mark Gerstein21, Morgan C. Giddings10, Thomas R. Gingeras4,38, Eric D. Green34, Roderic Guigo22,39, Ross C. Hardison25,26, Timothy J. Hubbard8, Manolis Kellis19, W. James Kent15, Jason D. Lieb18, Elliott H. Margulies31{, Richard M. 14 12 Myers , Michael Snyder , John A. Stamatoyannopoulos40, Scott A. Tenenbaum5, Zhiping Weng23, Kevin P. White41, Barbara Wold29,42; Boise State University and University of North Carolina at Chapel Hill Proteomics groups (data production and analysis) Jainab Khatun10, Yanbao Yu43, John Wrobel10, Brian A. Risk10, Harsha P. Gunawardena43, Heather C. Kuiper43, Christopher W. Maier43, Ling Xie43, Xian Chen43, Morgan C. Giddings10; Broad Institute Group (data production and analysis) Bradley E. Bernstein6,33, Charles B. Epstein6, Noam Shoresh6, Jason Ernst19{, Pouya Kheradpour19, Tarjei S. Mikkelsen6, Shawn Gillespie33, Alon Goren6,33, Oren Ram6,33, Xiaolan Zhang6, Li Wang6, Robbyn Issner6, Michael J. Coyne6, Timothy Durham6, Manching Ku6,33, Thanh Truong6, Lucas D. Ward19, Robert C. Altshuler19, Matthew L. Eaton19, Manolis Kellis19; Cold Spring Harbor, University of Geneva, Center for Genomic Regulation, Barcelona, RIKEN, Sanger Institute, University of Lausanne, Genome Institute of Singapore group (data production and analysis) Sarah Djebali22, Carrie A. Davis4, Angelika Merkel22, Alex Dobin4, Timo Lassmann28, Ali Mortazavi30, Andrea Tanzer22, Julien Lagarde22, Wei Lin4, Felix Schlesinger4, Chenghai Xue4, Georgi K. Marinov29, Jainab Khatun10, Brian A. Williams29, Chris Zaleski4, Joel Rozowsky21, Maik Roder22, Felix Kokocinski8{, Rehab F. Abdelhamid28, Tyler Alioto22,44, Igor Antoshechkin29, Michael T. Baer4, Philippe Batut4, Ian Bell45, Kimberly Bell4, Sudipto Chakrabortty4, Xian Chen43, Jacqueline Chrast46, Joao Curado22, Thomas Derrien22{, Jorg Drenkow4, Erica Dumais45, Jackie Dumais45, Radha Duttagupta45, Megan Fastuca4, Kata Fejes-Toth4{, Pedro Ferreira22, Sylvain Foissac45, Melissa J. Fullwood47{, Hui Gao45, David Gonzalez22, Assaf Gordon4, Harsha P. Gunawardena43, Cedric Howald46, Sonali Jha4, Rory Johnson22, Philipp Kapranov45{, Brandon King29, Colin 22,44 48 47 30 Kingswood , Guoliang Li , Oscar J. Luo , Eddie Park , Jonathan B. Preall4, Kimberly Presaud4, Paolo Ribeca22,44, Brian A. Risk10, Daniel Robyr49, Xiaoan Ruan47, Michael Sammeth22,44, Kuljeet Singh Sandhu47, Lorain Schaeffer29, Lei-Hoon See4, Atif Shahab47, Jorgen Skancke22, Ana Maria Suzuki28, Hazuki Takahashi28, Hagen Tilgner22{, Diane Trout29, Nathalie Walters46, Huaien Wang4, John Wrobel10, Yanbao Yu43, Yoshihide Hayashizaki28, Jennifer Harrow8, Mark Gerstein21, Timothy J. Hubbard8, Alexandre Reymond46, Stylianos E. Antonarakis49, Gregory J. Hannon4, Morgan C. Giddings10, Yijun Ruan47, Barbara Wold29,42, Piero Carninci28, Roderic Guigo22,39, Thomas R. Gingeras4,38; Data coordination center at UC Santa Cruz (production data coordination) Kate R. Rosenbloom15, Cricket A. Sloan15, Katrina Learned15, Venkat S. Malladi15, Matthew C. Wong15, Galt P. Barber15, Melissa S. Cline15, Timothy R. Dreszer15, Steven G. Heitner15, Donna Karolchik15, W. James Kent15, Vanessa M. Kirkup15, Laurence R. Meyer15, Jeffrey C. Long15, Morgan Maddren15, Brian J. Raney15; Duke University, EBI, University of Texas, Austin, University of North Carolina-Chapel Hill group (data production and analysis) Terrence S. Furey24, Lingyun Song17, Linda L. Grasfeder18, Paul G. Giresi18, Bum-Kyu Lee13, Anna Battenhouse13, Nathan C. Sheffield17, Jeremy M. Simon18, Kimberly A. Showers18, Alexias Safi17, Darin London17, Akshay A. Bhinge13, Christopher Shestak18, Matthew R. Schaner18, Seul Ki Kim18, Zhuzhu Z. Zhang18, Piotr A. Mieczkowski50, Joanna O. Mieczkowska18, Zheng Liu13, Ryan M. McDaniell13, Yunyun Ni13, Naim U. Rashid51, Min Jae Kim18, Sheera Adar18, Zhancheng Zhang24, Tianyuan Wang17, Deborah Winter17, Damian Keefe1, Ewan Birney1, Vishwanath R. Iyer13, Jason D. Lieb18, Gregory E. Crawford17,36; Genome Institute of Singapore group (data production and analysis) Guoliang Li48, Kuljeet Singh Sandhu47, Meizhen Zheng47, Ping Wang47, Oscar J. Luo47, Atif Shahab47, Melissa J. Fullwood47{, Xiaoan Ruan47, Yijun Ruan47; HudsonAlpha Institute, Caltech, UC Irvine, Stanford group (data production and analysis) Richard M. Myers14, Florencia Pauli14, Brian A. Williams29, Jason Gertz14, Georgi K. Marinov29, Timothy E. Reddy14{, Jost Vielmetter29,42, E. Christopher Partridge14, Diane Trout29, Katherine E. Varley14, Clarke Gasper29,42, Anita Bansal14, Shirley Pepke29,52, Preti Jain14, Henry Amrhein29, Kevin M. Bowling14, Michael Anaya29,42, Marie K. Cross14, Brandon King29, Michael A. Muratet14, Igor Antoshechkin29, Kimberly M. Newberry14, Kenneth McCue29, Amy S. Nesmith14, Katherine I. Fisher-Aylor29,42, Barbara Pusey14, Gilberto DeSalvo29,42, Stephanie L. Parker14{, Sreeram Balasubramanian29,42, Nicholas S. Davis14, Sarah K. Meadows14, Tracy Eggleston14, Chris Gunter14, J. Scott Newberry14, Shawn E. Levy14, Devin M. Absher14, Ali Mortazavi30, Wing H. Wong53, Barbara Wold29,42; Lawrence Berkeley National Laboratory group (targeted experimental validation) Matthew J. Blow54, Axel Visel54,55, Len A. Pennachio54,55; NHGRI groups (data production and analysis) Laura Elnitski37, Elliott H. Margulies31{, Stephen C. J. Parker31, Hanna M. Petrykowska37; Sanger Institute, Washington University, Yale University, Center for Genomic Regulation, Barcelona, UCSC, MIT, University of Lausanne, CNIO group (data production and analysis) Alexej Abyzov21, Bronwen Aken8, Daniel Barrell8, Gemma Barson8, Andrew Berry8, Alexandra Bignell8, Veronika Boychenko8, Giovanni Bussotti22, Jacqueline Chrast46, Claire Davidson8, Thomas Derrien22{, Gloria Despacio-Reyes8, Mark Diekhans15, Iakes Ezkurdia56, Adam Frankish8, James Gilbert8, Jose Manuel Gonzalez8, Ed Griffiths8, Rachel Harte15, David A. Hendrix19, Cedric Howald46, Toby Hunt8, Irwin Jungreis19, Mike Kay8, Ekta Khurana21, Felix Kokocinski8{, Jing Leng21, Michael F. Lin19, Jane Loveland8, Zhi Lu57, Deepa Manthravadi8, Marco Mariotti22, Jonathan Mudge8, Gaurab Mukherjee8, Cedric Notredame22, Baikang Pei21, Jose Manuel Rodriguez56, Gary Saunders8, Andrea Sboner58, Stephen Searle8, Cristina Sisu21, Catherine Snow8, Charlie Steward8, Andrea Tanzer22, Electra Tapanari8, Michael L. Tress56, Marijke J. van Baren59{, Nathalie Walters46, Stefan Washietl19, Laurens Wilming8, Amonida Zadissa8, Zhengdong Zhang60, Michael Brent59, David Haussler61, Manolis Kellis19, Alfonso Valencia56, Mark Gerstein21, Alexandre Reymond46, Roderic Guigo22,39, Jennifer Harrow8, Timothy J. Hubbard8; Stanford-Yale, Harvard, University of Massachusetts Medical School, University of Southern California/UC Davis group (data production and analysis) Stephen G. Landt12, Seth Frietze7, Alexej Abyzov21, Nick Addleman12, Roger P. Alexander21, Raymond K. Auerbach21, Suganthi Balasubramanian21, Keith Bettinger12, Nitin Bhardwaj21, Alan P. Boyle12, Alina R. Cao62, Philip Cayting12, Alexandra Charos63, Yong Cheng12, Chao Cheng21, Catharine Eastman12, Ghia Euskirchen12, Joseph D. Fleming64, Fabian Grubert12, Lukas Habegger21, Manoj Hariharan12, Arif Harmanci21, Sushma Iyengar65, Victor X. Jin66, Konrad J. Karczewski12, Maya Kasowski12, Phil Lacroute12, Hugo Lam12, Nathan Lamarre-Vincent64, Jing Leng21, Jin Lian67, Marianne Lindahl-Allen64, Renqiang Min21{, Benoit Miotto64, Hannah Monahan63, Zarmik Moqtaderi64, Xinmeng J. Mu21, Henriette OGeen62, Zhengqing Ouyang12, Dorrelyn Patacsil12, Baikang Pei21, Debasish Raha63, Lucia Ramirez12, Brian Reed63, Joel Rozowsky21, Andrea Sboner58, Minyi Shi12, Cristina Sisu21, Teri Slifer12, Heather Witt7, Linfeng Wu12, Xiaoqin Xu62, Koon-Kiu Yan21, Xinqiong Yang12, Kevin Y. Yip21{, Zhengdong Zhang60, Kevin Struhl64, Sherman M. Weissman67, Mark Gerstein21, Peggy J. Farnham7, Michael Snyder12; University of Albany SUNY group (data production and analysis) Scott A. Tenenbaum5, Luiz O. Penalva68, Francis Doyle5; University of Chicago, Stanford group (data production and analysis) Subhradip Karmakar41, Stephen G. Landt12, Raj R. Bhanvadia41, Alina Choudhury41, Marc Domanus41, Lijia Ma41, Jennifer Moran41, Dorrelyn Patacsil12, Teri Slifer12, Alec Victorsen41, Xinqiong Yang12, Michael Snyder12, Kevin P. White41; University of Heidelberg group (targeted experimental validation) Thomas Auer69{, Lazaro Centanin69, Michael Eichenlaub69, Franziska Gruhl69, Stephan Heermann69, Burkhard Hoeckendorf69, Daigo Inoue69, Tanja Kellner69, Stephan Kirchmaier69, Claudia Mueller69, Robert Reinhardt69, Lea Schertel69, Stephanie Schneider69, Rebecca Sinn69, Beate Wittbrodt69, Jochen Wittbrodt69; University of Massachusetts Medical School Bioinformatics group (data production and analysis) Zhiping Weng23, Troy W. Whitfield23, Jie Wang23, Patrick J. Collins3, Shelley F. Aldred3, Nathan D. Trinklein3, E. Christopher Partridge14, Richard M. Myers14; University of Massachusetts Medical School Genome Folding group (data production and analysis) Job Dekker11, Gaurav Jain11, Bryan R. Lajoie11, Amartya Sanyal11; University of Washington, University of Massachusetts Medical Center group (data production and analysis) Gayathri Balasundaram70, Daniel L. Bates16, Rachel Byron70, Theresa K. Canfield16, Morgan J. Diegel16, Douglas Dunn16, Abigail K. Ebersol71, Tristan Frum71, Kavita Garg72, Erica Gist16, R. Scott Hansen71, Lisa Boatman71, Eric Haugen16, Richard Humbert16, Gaurav Jain11, Audra K. Johnson16, Ericka M. Johnson71, Tattyana V. Kutyavin16, Bryan R. Lajoie11, Kristen Lee16, Dimitra Lotakis71, Matthew T. Maurano16, Shane J. Neph16, Fiedencio V. Neri16, Eric D. Nguyen71, Hongzhu Qu16, Alex P. Reynolds16, Vaughn Roach16, Eric Rynes16, Peter Sabo16, Minerva E. Sanchez71, Richard S. Sandstrom16, Amartya Sanyal11, Anthony O. Shafer16, Andrew B. Stergachis16, Sean Thomas16, Robert E. Thurman16, Benjamin Vernot16, Jeff Vierstra16, Shinny Vong16, Hao Wang16, Molly A. Weaver16, Yongqi Yan71, Miaohua Zhang70, Joshua M. Akey16, Michael Bender70, Michael O. Dorschner73, Mark Groudine70, Michael J. MacCoss16, Patrick Navas71, George Stamatoyannopoulos71, Rajinder Kaul9, Job Dekker11, John A. Stamatoyannopoulos40; Data Analysis Center (data analysis); Ian Dunham1, Kathryn Beal1, Alvis Brazma74, Paul Flicek1, Javier Herrero1, Nathan Johnson1, Damian Keefe1, Margus Lukk74{, Nicholas M. Luscombe75, Daniel Sobral1{, Juan M. Vaquerizas75, Steven P. Wilder1, Serafim Batzoglou2, Arend Sidow76, Nadine Hussami2, Sofia Kyriazopoulou-Panagiotopoulou2, Max W. Libbrecht2{, Marc A. Schaub2, Anshul Kundaje2{, Ross C. Hardison25,26, Webb Miller25, Belinda Giardine25, Robert S. Harris25, Weisheng Wu25, Peter J. Bickel20, Balazs Banfai20, Nathan P. Boley20, James B. Brown20, Haiyan Huang20, Qunhua Li20{, Jingyi Jessica Li20, William Stafford Noble16,77, Jeffrey A. Bilmes78, Orion J. Buske16, Michael M. Hoffman16, Avinash D. Sahu16{, Peter V. Kharchenko79, Peter J. Park79, Dannon Baker80, James Taylor80, Zhiping Weng23, Sowmya Iyer27, Xianjun Dong23, Melissa Greven23, Xinying Lin23, Jie Wang23, Hualin S. Xi32, Jiali Zhuang23, Mark Gerstein21, Roger P. Alexander21, Suganthi Balasubramanian21, Chao Cheng21, Arif Harmanci21, Lucas Lochovsky21, Renqiang Min21{, Xinmeng J. Mu21, Joel Rozowsky21, Koon-Kiu Yan21, Kevin Y. Yip21{ & Ewan Birney1
Vertebrate Genomics Group, European Bioinformatics Institute (EMBL-EBI), Wellcome Trust Genome Campus, Hinxton, Cambridgeshire CB10 1SD, UK. 2Department of Computer Science, Stanford University, 318 Campus Drive, Stanford, California 94305-5428, USA. 3SwitchGear Genomics, 1455 Adams Drive Suite 1317, Menlo Park, California 94025, USA. 4Functional Genomics, Cold Spring Harbor Laboratory, 1 Bungtown Road, Cold Spring Harbor, New York 11724, USA. 5College of Nanoscale Sciences and Engineering, University ay Albany-SUNY, 257 Fuller Road, NFE 4405, Albany, New York 12203, USA. 6Broad Institute of MIT and Harvard, 7 Cambridge Center, Cambridge, Massachusetts 02142, USA. 7Biochemistry and Molecular Biology, USC/ Norris Comprehensive Cancer Center, 1450 Biggy Street, NRT 6503, Los Angeles, California 90089, USA. 8Informatics, Wellcome Trust Sanger Institute, Wellcome Trust Genome Campus, Hinxton, Cambridgeshire CB10 1SA, UK. 9Department of Medicine, Division of Medical Genetics, University of Washington, 3720 15th Avenue NE, Seattle, Washington 98195, USA. 10College of Arts and Sciences, Boise State University, 1910 University Drive, Boise, Idaho 83725, USA. 11Program in Systems Biology, Program in Gene Function and Expression, Department of Biochemistry and Molecular
1
RESEARCH ARTICLE
Pharmacology, University of Massachusetts Medical School, 364 Plantation Street, Worcester, Massachusetts 01605, USA. 12Department of Genetics, Stanford University, 300 Pasteur Drive, M-344, Stanford, California 94305-5120, USA. 13Center for Systems and Synthetic Biology, Institute for Cellular and Molecular Biology, Section of Molecular Genetics and Microbiology, The University of Texas at Austin, 1 University Station A4800, Austin, Texas 78712, USA. 14HudsonAlpha Institute for Biotechnology, 601 Genome Way, Huntsville, Alabama 35806, USA. 15Center for Biomolecular Science and Engineering, University of California, Santa Cruz, 1156 High Street, Santa Cruz, California 95064, USA. 16 Department of Genome Sciences, University of Washington, 3720 15th Ave NE, Seattle, Washington 98195-5065, USA. 17Institute for Genome Sciences and Policy, Duke University, 101 Science Drive, Durham, North Carolina 27708, USA. 18Department of Biology, Carolina Center for Genome Sciences, and Lineberger Comprehensive Cancer Center, The University of North Carolina at Chapel Hill, 408 Fordham Hall, Chapel Hill, North Carolina 27599-3280, USA. 19Computer Science and Artificial Intelligence Laboratory, Broad Institute of MIT and Harvard, Massachusetts Institute of Technology, 32 Vassar Street, Cambridge, Massachusetts 02139, USA. 20Department of Statistics, University of California, Berkeley, 367 Evans Hall, University of California, Berkeley, Berkeley, California 94720, USA. 21Computational Biology and Bioinformatics Program, Yale University, 266 Whitney Avenue, New Haven, Connecticut 06520, USA. 22 Bioinformatics and Genomics, Centre for Genomic Regulation (CRG) and UPF, Doctor Aiguader, 88, Barcelona 08003, Catalonia, Spain. 23Program in Bioinformatics and Integrative Biology, University of Massachusetts Medical School, 364 Plantation Street, Worcester, Massachusetts 01605, USA. 24Department of Genetics, The University of North Carolina at Chapel Hill, 120 Mason Farm Road, CB 7240, Chapel Hill, North Carolina 27599, USA. 25Center for Comparative Genomics and Bioinformatics, The Pennsylvania State University, Wartik Laboratory, University Park, Pennsylvania 16802, USA. 26 Department of Biochemistry and Molecular Biology, The Pennsylvania State University, 304 Wartik Laboratory, University Park, Pennsylvania 16802, USA. 27Program in Bioinformatics, Boston University, 24 Cummington Street, Boston, Massachusetts 02215, USA. 28RIKEN Omics Science Center, RIKEN Yokohama Institute, 1-7-22 Suehiro-cho, Tsurumi-ku, Yokohama, Kanagawa 230-0045, Japan. 29Division of Biology, California Institute of Technology, 156-291200 East California Boulevard, Pasadena, California 91125, USA. 30Developmental and Cell Biology and Center for Complex Biological Systems, University of California Irvine, 2218 Biological Sciences III, Irvine, California 92697-2300, USA. 31Genome Technology Branch, National Human Genome Research Institute, 5625 Fishers Lane, Bethesda, Maryland 20892, USA. 32Department of Biochemistry and Molecular Pharmacology, Bioinformatics Core, University of Massachusetts Medical School, 364 Plantation Street, Worcester, Massachusetts 01605, USA. 33Howard Hughes Medical Institute and Department of Pathology, Massachusetts General Hospital and Harvard Medical School, 185 Cambridge St CPZN 8400, Boston, Massachusetts 02114, USA. 34National Human Genome Research Institute, National Institutes of Health, 31 Center Drive, Building 31, Room 4B09, Bethesda, Maryland 20892-2152, USA. 35National Human Genome Research Institute, National Institutes of Health, 5635 Fishers Lane, Bethesda, Maryland 20892-9307, USA. 36Department of Pediatrics, Division of Medical Genetics, Duke University School of Medicine, Durham, North Carolina 27710, USA. 37National Human Genome Research Institute, National Institutes of Health, 5625 Fishers Lane, Rockville, Maryland 20892, USA. 38Affymetrix, Inc., 3380 Central Expressway, Santa Clara, California 95051, USA. 39Departament de `ncies Experimentals i de la Salut, Universitat Pompeu Fabra, Barcelona, Catalonia Cie 08002, Spain. 40Department of Genome Sciences, Box 355065, and Department of Medicine, Division of Oncology, Box 358081, University of Washington, Seattle, Washington 98195-5065, USA. 41Institute for Genomics and Systems Biology, The University of Chicago, 900 East 57th Street, 10100 KCBD, Chicago, Illinois 60637, USA. 42 Beckman Institute, California Institute of Technology, 156-29 1200 E. California Boulevard, Pasadena, California 91125, USA. 43Department of Biochemistry and Biophysics, University of North Carolina School of Medicine, Campus Box 7260, 120 Mason Farm Road, 3010 Genetic Medicine Building, Chapel Hill, North Carolina 27599, lisis Geno mico (CNAG), C/Baldiri Reixac 4, Torre I, USA. 44Centro Nacional de Ana Barcelona, Catalonia 08028, Spain. 45Genomics, Affymetrix, Inc., 3380 Central Expressway, Santa Clara, California 95051, USA. 46Center for Integrative Genomics, University of Lausanne, Genopode Building, 1015 Lausanne, Switzerland. 47Genome Technology and Biology, Genome Institute of Singapore, 60 Biopolis Street, 02-01, Genome, Singapore 138672, Singapore. 48Computational and Systems Biology, Genome Institute of Singapore, 60 Biopolis Street, 02-01, Genome, Singapore 138672, Singapore. 49 Department of Genetic Medicine and Development, University of Geneva Medical School, and University Hospitals of Geneva, 1 rue Michel-Servet, 1211 Geneva 4, Switzerland. 50Department of Genetics, The University of North Carolina at Chapel Hill, 5078 GMB, Chapel Hill, North Carolina 27599-7264, USA. 51Department of Biostatistics, Gillings School of Global Public Health, The University of North Carolina at Chapel Hill, 408 Fordham Hall, Chapel Hill, North Carolina 27599-7445, USA. 52Center for Advanced Computing Research, California Institute of Technology, MC 158-79, 1200 East California Boulavard, Pasadena, California 91125, USA. 53Department of Statistics, Stanford University, Sequoia Hall. 390 Serra Mall, Stanford, California 94305-4065, USA. 54DOE Joint Genome Institute, Walnut Creek, California, USA. 55Genomics Division, Lawrence Berkeley National Laboratory, One Cyclotron Road, MS 84-171, Berkeley, California 94720, USA. 56Structural Computational Biology, Spanish National Cancer Research Centre (CNIO), Melchor Fernandez Almagro, 3, 28029 Madrid, Spain. 57School of Life Sciences, Tsinghua University, School of Life Sciences, Tsinghua University, 100084 Beijing, China. 58Department of Pathology and Laboratory Medicine, Institute for Computational Biomedicine, Weill Cornell Medical College, 1305 York Avenue, Box 140, New York, New York 10065, USA. 59Computer Science and Engineering, Washington University in St Louis, St Louis, Missouri 63130, USA. 60Department of Genetics, Albert Einstein College of Medicine, 1301 Morris Park Avenue, Room 353A, Bronx, New York 10461, USA. 61Center for Biomolecular Science and Engineering, Howard Hughes Medical Institute, University of California, Santa Cruz, 1156 High Street, Santa Cruz, California 95064, USA. 62Genome Center, University of California-Davis, 451 Health Sciences Drive, Davis, California 95616, USA. 63Department of Molecular, Cellular, and Developmental Biology, Yale University, 266 Whitney Avenue, New Haven, Connecticut 06511, USA. 64Biological Chemistry and Molecular Pharmacology, Harvard Medical School, 240 Longwood Avenue, Boston, Massachusetts 02115, USA. 65Biochemistry and Molecular Biology, University of Southern California, 1501 San Pablo Street, Los Angeles, California 90089, USA. 66Department of Biomedical Informatics, Ohio State University, 3172C Graves Hall, 333 W Tenth Avenue, Columbus, Ohio 43210, USA. 67Department of Genetics, Yale University, Yale University School of Medicine, 333 Cedar Street, New Haven, Connecticut 06510, USA. 68Department of Cellular and Structural Biology, Childrens Cancer Research InstituteUTHSCSA, Mail code 7784- 7703 Floyd Curl Dr, San Antonio, Texas 78229, USA. 69Centre for Organismal Studies (COS) Heidelberg, University of Heidelberg, Im Neuenheimer Feld 230, 69120 Heidelberg, Germany. 70Basic Sciences Division, Fred Hutchinson Cancer Research Center, 825 Eastlake Avenue East, Seattle, Washington 98109, USA. 71Department of Medicine, Division of Medical Genetics, Box 357720, University of Washington, Seattle, Washington 98195-7720, USA. 72Division of Human Biology, Fred Hutchinson Cancer Research Center, 825 Eastlake Avenue East, Seattle, Washington 98109, USA. 73Department of Psychiatry and Behavioral Sciences, Box 356560, University of Washington, Seattle, Washington 98195-6560, USA. 74 Microarray Informatics Group, European Bioinformatics Institute (EMBL-EBI), Wellcome Trust Genome Campus, Hinxton, Cambridgeshire CB10 1SD, UK. 75Genomics and Regulatory Systems Group, European Bioinformatics Institute (EMBL-EBI), Wellcome Trust Genome Campus, Hinxton, Cambridgeshire CB10 1SD, UK. 76Department of Pathology, Department of Genetics, Stanford University, 300 Pasteur Drive, Stanford, California 94305, USA. 77Department of Computer Science and Engineering, 185 Stevens Way, Seattle, Washington 98195, USA. 78Department of Electrical Engineering, University of Washington, 185 Stevens Way, Seattle, Washington 98195, USA. 79Center for Biomedical Informatics, Harvard Medical School, 10 Shattuck Street, Boston, Massachusetts 02115, USA. 80Departments of Biology and Mathematics and Computer Science, Emory University, Atlanta, Georgia 30322, USA. {Present addresses: Computer Science and Artificial Intelligence Laboratory, Massachusetts Institute of Technology, 32 Vassar Street, Cambridge, Massachusetts 02139, USA (A.K.); UCLA Biological Chemistry Department, Eli and Edythe Broad Center of Regenerative Medicine and Stem Cell Research at UCLA, Jonsson Comprehensive Cancer Center, 615 Charles E Young Dr South, Los Angeles, California 90095, USA (J.E.); Department of Statistics, 514D Wartik Lab, Penn State University, State College, Pennsylvania 16802, USA (Q.L.); Department of Biostatistics and Bioinformatics and the Institute for Genome Sciences and Policy, Duke University School of Medicine, 101 Science Drive, Durham, North Carolina 27708, USA (T.E.R.); Department of Computer Science and Engineering, The Chinese University of Hong Kong, Shatin, New Territories, Hong Kong (K.Y.Y.); Department of Genetics, Washington University in St Louis, St Louis, Missouri 63110, USA (R.F.L.); Department of Cell Biology and Molecular Genetics, University of Maryland, College Park, Maryland 20742, USA (L.A.L.D.); National Cancer Institute, National Institutes of Health, Bethesda, Maryland 20892, USA (J.Z.); University of California, Davis Population Biology Graduate Group, Davis, California 95616, USA (J.R.W.); Illumina Cambridge Ltd., Chesterford Research Park, Little Chesterford, Saffron Walden, Essex CB10 1XL, UK (E.H.M.); BlueGnome Ltd., CPC4, Capital Park, Fulbourn, Cambridge CB21 5XE, UK (F.K.); Institut de Genetique et Developpement de Rennes, CNRS-UMR6061, Universite de Rennes 1, F-35000 Rennes, Brittany, France (T.D.); Caltech, 1200 East California Boulevard, Pasadena, California 91125, USA (K.F.-T.); A*STAR-Duke-NUS Neuroscience Research Partnership, 8 College Road, Singapore 169857, Singapore (M.J.F.); St Laurent Institute, One Kendall Square, Cambridge, Massachusetts 02139, USA (P.K.); Department of Genetics, Stanford University, Stanford, California 94305, USA (H.T.); Biomedical Sciences (BMS) Graduate Program, University of California, San Francisco, 513 Parnassus Avenue, HSE-1285, San Francisco, California 94143-0505, USA (S.L.P.); Monterey Bay Aquarium Research Institute, Moss Landing, California 95039, USA (M.J.v.B.); Department of Machine Learning, NEC Laboratories America, 4 Independence Way, Princeton, New Jersey 08540, USA (R.M.); Neuronal Circuit Development Group, Unite de Genetique et Biologie du Developpement, U934/UMR3215, Institut Curie Centre de Recherche, Pole de Biologie du Developpement et Cancer, 26, rue dUlm, 75248 Paris Cedex 05, France (T.A.); Cancer Research UK, Cambridge Research Institute, Li Ka Shing Centre, Robinson Way, Cambridge CB2 0RE, UK (M.L.); Unidade de Bioinformatica, Rua da Quinta Grande, 6, P-2780-156 Oeiras, Portugal (D.S.); Department of Genome Sciences, University of Washington, 3720 15th Avenue NE, Seattle, Washington 98195-5065, USA (M.W.L.); Center for Bioinformatics and Computational Biology, 3115 Ag/Life Surge Building 296, University of Maryland, College Park, Maryland 20742, USA (A.D.S.).

Nature 11247

Uploaded by

Copyright:

Available Formats

Nature 11247

Uploaded by

Document Information

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Nature 11247

Uploaded by

Copyright:

Available Formats

ARTICLE

An integrated encyclopedia of DNA elements in the human genome

2012 Macmillan Publishers Limited. All rights reserved

2012 Macmillan Publishers Limited. All rights reserved

ChromRem DNARep HISase Other Pol2 Pol3 TFNS TFSS

2012 Macmillan Publishers Limited. All rights reserved

2012 Macmillan Publishers Limited. All rights reserved

Human and mammalian constraint at TSS-distal non-exonic DNase peaks

Human and mammalian constraint at TSS-distal non-exonic ENCODE elements

Human diversity (inverted scale)

Human diversity (inverted scale)

AP2 MYC EBF

Human diversity (inverted scale)

Glucocorticoid receptor transcription factor

Human diversity (inverted scale)

Primate-specific TSS-distal non-exonic regions GERP conservation score

0.20 Non-ENCODE ENCODE-annotated 0.10

0.00 1.0 0.5 0.0 0.5

ENCODE data integration with known genomic features

2012 Macmillan Publishers Limited. All rights reserved

CAGE poly(A)+ K562 whole cell

Pearsons r = 0.9 RMSE = 1.9 10

Regression (mean decreased Gini)

5 4 2 0 2 4 Predicted expression (log2) 6 8

Measured expression (log2)

5 Regression (mean decreased Gini)

5 4 2 0 2 4 Predicted expression (log2) 6 8

2012 Macmillan Publishers Limited. All rights reserved

YY M 1 Y EL C F E2 1 F JU 4 ET N RE S1 S M T G XI AB 1 P M A A FO X N S F EGYB R E2 1 ZN F F 6 TH 26 AP 3 S 1 N P2 RF TA 1 L ZBNFY1 TB A G 7A AT A SP 2 SI I1 X5 U FOSF2 SL S 1 G P1 AT BCA1 L BCUSF 3 LA 1 JU F1 ZB N TB D N 33 F M E2 AF K C SR EB F PB A N TF R 3 ZN 2C F2 2 74

YY M 1 Y M C RE AX ST ET S E2 1 F EL 4 EG F1 R1 TA L JU 1 E2 N BC F LA 6 F M 1 XI 1 G SP ZBABP1 TB A B 7A TH CL AP 3 S 1 G RF AT U A2 S N F2 FOFYA S M L1 ZB AF TB K 3 FO 3 N S ZN RF F2 1 U 63 G SF1 AT A SP 1 AT 2 C F E 3 N BPB R2 JU C2 N SP D N I1 FY B ZN SIX F2 5 N 74 FE 2

3K 9 3K ac 4 3K me 36 3 H me3 3K 27 DN ac H ase 3K I 4m e H 2 N H3K 2A. or Z m 27 al m ize e3 d H Cp 3K G 4 H me 3K 1 9 H me 3K 1 H 9m 4K e 20 3 m e C 1 on tro l H H

H 3K 9 3K ac 4 H me 3K 3 4m DN e2 a H se 3 I H K27 3K a 79 c m e2 H H2A 3K 36 .Z H 3K me 27 3 N H3 me or m K4 3 al m ize e1 d H Cp 3K G 9 H me 3K 1 H 9m 4K e 20 3 m e C 1 on tro l

Relative importance of variables

Distance from summit

Distance from summit

Nucleosome CTCF, CTCFL, RAD21, SMC3, ZNF143

2012 Macmillan Publishers Limited. All rights reserved

JUNB JUND JUN FOSL1

HDAC2 NANOG EP300 SP1

JUNB JUND JUN FOSL1

HDAC2 GABPA CHD2 POLR2A GTF2F1 MXI1 MYC

HDAC2 NANOG EP300 SP1

Degree of co-association (z-score)

Light red Grey

Dark green Yellow

2012 Macmillan Publishers Limited. All rights reserved

UW DNase Open chrom DNase FAIRE H3K4me1

GM12878 segment overlaps with transcription factors (obs./exp. coverage)

GM12878 overlaps with RNAs (obs./exp.)

Long RNA cytoplasm A+ Long RNA cytoplasm A

Enriched (log2 = 5.8)

Per cent methylated

Cell type in each CTCF segment

Cell type in each E segment

Cell type in each TSS segment

Cell type in each R segment

Distribution of methylation over GM12878 segment labels

Mean cell type in segment

2012 Macmillan Publishers Limited. All rights reserved

Depleted (log2 = 10.0)