Direct from the source! Thank you sjneph, much appreciated.
Hello.
I would like to download the ENCODE DNase I footprint data for the 41 human cell and tissue types described by Neph et al. (2012) Nature, 489:83-90.
Here is an example to illustrate my question. Consider only a single human cell type, K562. The following DNase I data files can be found via NCBI's GEO repository, using accession number GSE26328 (see http://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSM646567)
SRX/SRX037/SRX037116
GSM646567_hg19_wgEncodeUwDgfK562Aln.bam
GSM646567_hg19_wgEncodeUwDgfK562Hotspots.broadPeak.txt.gz
GSM646567_hg19_wgEncodeUwDgfK562Pk.narrowPeak.txt.gz
GSM646567_hg19_wgEncodeUwDgfK562Raw.bigWig
GSM646567_hg19_wgEncodeUwDgfK562Sig.bigWig
Upon downloading these files, it becomes apparent that none of them contain the actual footprints. Perhaps the most relevant file is GSM646567_hg19_wgEncodeUwDgfK562Hotspots.broadPeak.txt.gz, which contains 256,735 lines, each pertaining to a DNase I Hypersensitive Zone (see http://genome.ucsc.edu/cgi-bin/hgTrackUi?db=hg19&g=wgEncodeUwDgf). Indeed, this number matches up exactly with the number of "Hotspot Regions" reported in Table S1 of the Neph et al. (2012) paper.
What I am looking for, however, is a .bed file that lists the coordinates of the footprints that these Hotspot Regions contain. Based on Table S1, this file should have 498,683 lines (see "Number of Footprints" column).
Does anyone know where to find these files? Moreover, does anyone know where to find the Footprint Occupancy Scores for these 41 cell and tissue types?
2 answers
Look for DNAse digital footprinting files here: https://genome.ucsc.edu/cgi-bin/hgFileUi?db=hg19&g=wgEncodeUwDgf
I think the narrowPeak files contain some sort of occupancy score.
Thanks Mikael, but the link you provided is just a mirror of the link I posted.
It's unclear what the narrowPeak files contain. One would think they would contain the data summarized in the "Hotspot Regions (FDR 1%)" column of Table S1 in Neph et al. (2012). However, the K562 narrowPeak file contains 187,374 lines, while the corresponding entry in Table S1 indicates that there should be 125,859 lines.
Regardless, these peaks are much wider than the footprints, which should be between 6-40 bp long.
So, the question remains: Where can the ENCODE DNase I footprints be found?
Sorry, I didn't realize I was copying your link. I will delete this answer. Probably the discrepancy in numbers can partially explained by the fact that they have released several new versions with new peak calls.
Log in to answer this question.