For this analysis, we only use ENCODE data from the January 2011 freeze. ChIP-seq peaks determined using SPP (Kharchenko et al. 2008) at a loose threshold and a cutoff was selected based on the IDR method (A Kundaje, Q Li, B Brown, J Rozowsky, A Harmanci, S Wilder, S Batzoglou, I Dunham, M Gerstein, E Birney, et al., in prep.). DNase-seq data were based on “peak” calls from both Duke University and the University of Washington (this excludes hotspots).
