Epimap Data Exploration

Representative track hubs:

1. Full track hubs (all 17,938 tracks): WUSTL / UCSC
These full hubs are huge track hubs and will be very slow to load.
We recommend making custom hubs (next tab) to restrict to datasets of interest.

2. Average tracks for 33 major sample groups in each mark: WUSTL / UCSC

3. Chromatin state tracks: Epilogos / UCSC (hg19) / UCSC (hg38)

4. All epigenomes and average tracks for a specific sample group:

Selected sample group:

TrackHub Options:

Track type(s):

Imputed

Observed

Both

Marks/Assays (by set):

Tier 1 (Core Marks + DNase-seq)

Tier 2 (Secondary Marks + ATAC-seq)

Tier 3 (DNA Factors)

Tier 4 (Other Marks)

Specific histone marks and assays to add to the track hub:

Note: Some sample by assay combinations may either have both imputed and observed data or neither.

Download WUSTL TrackHub Download UCSC TrackHub Download File List

Usage Instructions:

WUSTL Epigenome Browser (Legacy):

1. Select tracks and download WUSTL TrackHub (json formatted)
2. Go to the WUSTL Browser (legacy)
3. Select the Tracks > Custom Tracks > Add new tracks > Datahub by upload > Upload File

NOTE: We do not recommend loading more than 100 tracks in the same WUSTL legacy track hub.
The WUSTL legacy client does not allow you to select specific tracks from a hub.

WUSTL Epigenome Browser (new):

1. Select tracks and download WUSTL TrackHub (json formatted)
2. Go to the WUSTL Browser (new)
3. Select the Tracks > Remote Tracks > Add Remote Data Hub > Upload File

The new WUSTL browser does allow you to select specific tracks from a hub, so your hub can be much larger than the legacy version, as long as you do not load everything in.
Do not load in the coordinate override or native tracks - these are for the legacy version.

UCSC Epigenome Browser:

1. Select tracks and download UCSC TrackHub (plain text, all in one file)
2. Host hub on a http or ftp server (or use GBiB to host UCSC locally)
3. Edit the following link with your hub location, then use it to connect the hub to the UCSC browser:

http://genome.ucsc.edu/cgi-bin/hgTracks?db=hg19&hubUrl=https://www.yourserver.org/hubfile.txt
We do not recommend loading more than 1000 tracks in the same UCSC track hub.
You do not have to display all tracks on the UCSC trackhub, but it may take a while to load in all of the data for many tracks.

Note: Select nodes to see edge and node descriptions below the network.

Network edges (for selected node):

Node summary (samples in cluster/motif logo):

Download Table

Note: Interactive heatmap (size = 833 x 300) may be slow to load.

Note: Interactive heatmap may be slow to load.

GWAS Enrichment on epigenomes hierarchy:

Tissue-specific enhancers near GWAS lead SNPs: All enhancers within 2.5kb of a GWAS lead SNP that are also active in one of the top-enriched tree nodes in the GWAS.

Download Enhancers

Table of gene-enhancer links: Gene-enhancer links in the GWAS loci (SNPs +/- 1Mb), reported for the top-enriched sample groups in the GWAS.

Download Links

Select locus:

Click to enable/disable zoom on locus, scroll to change zoom size.

Select multiple GWAS on sidebar to visualize their enrichments side by side:

About:

Main website: http://compbio.mit.edu/epimap/
Paper: https://www.nature.com/articles/s41586-020-03145-z
Contact: cboix@mit.edu

Tracks and Metadata

Sample metadata table (xlsx, tsv).
Track download (bigWig files hosted by WUSTL).
Average per-group and per-mark tracks (bigWig files hosted by the Broad).
Track hub files (WUSTL - json; and UCSC - txt).
Chromatin state calls using ChromHMM (hg19, hg38), calculated using the 18-state Roadmap model.

Enhancers, Modules, and Enrichments:

Enhancer locations: directory, enhancers (0-ind), promoters (0-ind); enh. matrix hdf5, mtx; prom. matrix hdf5, mtx
Enhancer modules: assignments, centers, and metadata directory
Enrichments: archetype motifs, full motifs, GO terms
Motif enrichments are on JASPAR 2018, HOCOMOCOv11, and HT-SELEX (Jolma, 2013).
Motifs are split into archetypes according to Vierstra (2020)

GWAS enrichments:

Note: We have updated the site to correct for two implementation mistakes in the GWAS enrichment analysis: 1. The first mistake was that we calculated the enrichment in the wrong direction: instead of calculating the enrichment of SNPs in biosample enhancers as was previously done in Roadmap, we mistakenly calculated the enrichment of enhancers in GWAS lead SNPs. This led to double-counting SNPs and inflated enrichment statistics in the case of multiple nearby enhancers within 2.5kb of the same lead SNP, which were all counted as independently overlapping the same SNP, when in fact the multiple overlaps came from their genomic clustering in the same locus. 2. The second mistake was that our FDR procedure only controlled for inflation due to multiple enhancers proximal to the same SNP, but not for the multiple hypothesis testing across many tissue-trait combinations, and thus our original reported statistics correspond to nominal P-values rather than FDR-corrected P-values.

We have updated the GWAS enrichments page to add a revised analysis that 1) tests for enrichment of SNPs in biosamples and 2) has appropriate FDR control (at 1%, 2%, and 5% FDR).

The resources for the revised analysis can be found here:

GWAS Enrichments (revised): tsv, and Rda
GWAS Figures: tree enrichments and SNP heatmaps and GWAS loci.
- All plotted links in GWAS loci are given per-folder as either tsv.gz or Rda files.
Enhancers near lead SNPs for enriched epigenomes: tsv or Rda
Directly linked genes: tsv or Rda

The resources for the original analysis can be found here:

GWAS Enrichments (original): tsv (1%), tsv (0.1%), and Rda
GWAS Figures: tree enrichments and SNP heatmaps and GWAS loci.
- All plotted links in GWAS loci are given per-folder as either tsv.gz or Rda files.
Enhancers near lead SNPs for enriched epigenomes: tsv or Rda
Directly linked genes: tsv or Rda

Linking:

For each sample: gene-enhancer links
Aggregated for each sample group (Adipose, Brain, etc.): gene-enhancer links