For each of the 39 Roadmap reference epigenomes with DNase data, peak positions are combined across reference epigenomes by defining peak island areas, defined by stacking all DNase peak positions across epigenomes, and considering the Full Width at Half Maximum (FWHM). Note that for this we are only considering peak locations, not intensities. The goal of this is to obtain an estimate of the area of open chromatin, not to quantify the level of ‘openness’, as these data are not available for all reference epigenomes. In cases when peak islands overlap, they are merged because it means that the original DNase peak area populations overlap at least for half of the epigenomes with DNase peaks in that area (given the FWHM approach). Peak island summits are defined as the median peak summit of all peak island member DNase peaks. This results in a total of 3,516,964 DNase enriched regions across epigenomes.
