The tracks contain chromatin-state maps generated at the Broad Institute using ChIP-Seq.
Sites by HMM
Enriched intervals inferred by an HMM
Sites by Windowing
Enriched intervals inferred by fixed-size windows
Signal Densities
ChIP-Seq fragment densities analyzed
Alignments
Sequences and coordinates of uniquely aligned reads
Sites by HMM
These subtracks contain the coordinates of intervals enriched
for H3K4Me3, H3K9Me3, H3K27Me3, H3K36Me3 or
H4K20Me3 in mouse ES cells, as determined by a
Hidden Markov Model. Prior to analysis, the fragment densities were
discretized into four classes (masked, sub-threshold, near-threshold,
above-threshold), which were defined for each sample by supervised
training on a 10 Mb subset of the genome.
Sites by Windowing
These subtracks contain the coordinates of intervals significantly
enriched for H3K4Me3, H3K4me2, H3K4me1 or H3K27Me3 in ES cells,
ES-derived neural progenitors (NP), whole brain tissue and embryonic fibroblasts (MEF).
Intervals were obtained by comparing the average densitity of a 1-kb
window to the cumulative density distribution expected from randomly
aligning reads across the unique portion of the reference
genome, and merging any overlapping windows with nominal p-values of < 10-5.
The coordinates indicate the beginning of the
first window and the end of the last window. The fourth column
indicates the maximal fragment density within each interval.
H3K9Me3 enriched sites have been added since the initial
publication of this data. These sites correspond to a somewhat more permissive threshold than the
HMM-based sites analyzed in the manuscript. The maximum fragment
density can be used to rank the sites by confidence.
Signal Densities
These subtracks contains the ChIP-Seq fragment densities
analyzed in the paper. They represent the density of antibody
enriched fragments in celltype at 25 bp resolution across the mouse genome.
A positive value indicates the number of uniquely aligned
fragments oriented towards, and within 300 bp of, a particular
position. Reads within 200 bp add 1 to the density, and reads
within 300 bp add 0.25. (This is an approximation of the expected
number of fragments overlapping the position, since the exact length
of each is unknown).
Alignments
These subtracks contain alignment coordinates of the ChIP-Seq reads on
the mouse reference genome (UCSC version mm8).
SAMPLE
ANTIBODIES
ES cells
Whole cell extract (unenriched input)
pan-H3
H3K4Me3
H3K4Me2
H3K4Me1
H3K27Me3
H3K36Me3
H3K9Me3
H4K20Me3
RNA polymerase II
ES hybrid cells (129/CAST)
H3K4Me3
H3K36Me3
H3K9Me3
MEF
Whole cell extract (unenriched input)
H3K4Me3
H3K27Me3
H3K36Me3
H3K9Me3
NP cells
Whole cell extract (unenriched input)
H3K4Me3
H3K4Me2
H3K4Me1
H3K27Me3
H3K36Me3
H3K9Me3
Whole brain
H3K4Me3
H3K4Me2
H3K27Me3
Methods
Growth protocol
V6.5 murine ES cells (genotype 129SvJae X C57BL/6; male; passages
10-15) and hybrid murine ES cells (genotype 129SvJae X
M. m. castaneus F1; male; passages 4-6) were cultivated in 5%
CO at 37C on irradiated MEFs in DMEM containing 15% FCS, LIF,
penicillin/streptomycin, L-glutamine, nonessential amino-acids and
2-mercaptoethanol. Cells were subjected to at least 2-3 passages on
0.2% gelatin under feeder-free conditions prior to analysis. V6.5 ES
cells were differentiated into neural progenitor (NP) cells through
embryoid body formation for 4 days and selection in ITSFn media for
5-7 days, and maintained by FGF2 and EGF2. Mouse embryonic fibroblasts
(genotype 129SvJae X C57BL/6; male; E13.5; passages 4-6) were grown in
DMEM with 10% fetal bovine serum and penicillin/streptomycin at 37C,
5% CO2. Primary tissues were isolated from 4-6 weeks old male
129SvJae/C57BL/6 mice.
Extraction protocol
Cells and homogenized tissues were fixed in 1% formaldehyde and
resuspended in lysis buffer. Chromatin was sheared to
200-700 bp using a Branson 250 Sonifier or Diagenode Bioruptor. Solubilized
chromatin was immunoprecipitated with antibodies against H3K4me3
(Abcam 8580), H3K9me3 (Abcam 8898), H3K27me3 (Upstate 07-449),
H3K36me3 (Abcam 9050), H4K20me3 (Upstate 07-463), pan-H3 (Abcam 1791)
or RNA polymerase II (Covance MMS-126R). Antibody-chromatin complexes
were pulled-down using protein A-sepharose (or anti-IgM-conjugated
agarose for RNA polymerase II), washed and then eluted. After
cross-link reversal and proteinase K treatment, immunoprecipitated DNA
was extracted with phenol-chloroform, ethanol precipitated, treated
with RNAse and purified. One to ten nanograms of DNA were
end-repaired, adapter-ligated and sequenced by Illumina Genome
Analyzers as recommended by the manufacturer.
Data processing
Sequence reads from each IP experiment were aligned to the mouse
reference genome (mm8) using the ARACHNE computational
pipeline. First, a table was pre-computed to associate all possible 12-mers with all of
their occurrences in the genome. For each ChIP-Seq read (forward
and reverse complement orientation), each potential start point was
then found and the number of mismatches in the corresponding gap-free
alignment was computed. All uniquely aligned reads (defined as the
second to best alignment having >2 mismatches more than the best
alignment, and the total mismatch count being <=6) were kept. If
multiple reads aligned to the same starting position, only one were
kept. Fragment densities were computed by counting the number of reads
(extended to 300 bp) overlapping each position in the genome (at 25 bp
resolution). Non-unique positions in the reference genome
were pre-computed by aligning every 27-mer in the genome to the whole
genome and masking positions that did not meet the uniqueness criteria
defined above.