Extended Data Figure 3: DNase I footprints accurately recapitulate ChIP-seq data.

For five different TFs with corresponding ChIP-seq data in MEL cells, displayed are (left) heat maps showing per-nucleotide DNase I cleavage and (right) vertebrate conservation by phyloP for all motif instances of that TF within MEL DNase I hotspots (irrespective of whether they overlap a DNase I footprint), ranked by the local density of DNase I cleavages. The number of motif instances for that TF is indicated to the left of the heat map. Purple ticks indicate the presence of the corresponding TF ChIP-seq peaks at each motif instance. Green ticks indicate the presence of DNase I footprints at each motif instance. Below each graph is indicated the percentage of TF footprints that reside outside of a ChIP-seq verified binding site, as well as the percentage of ChIP-seq peaks that do not contain a DNase I footprint for that TF (indicating indirect TF occupancy). Of note, occupied motifs within DNase I footprints accurately recapitulate sites of direct TF occupancy, as 99% of DNase I footprinted motifs for a given TF overlap a cognate ChIP-seq peak. In contrast, for most TFs the majority of ChIP-seq peaks arise from indirect TF occupancy events (and thus lack DNase I footprinted sequence elements for their cognate TF).