Fig. 3

Workflow of DiNeR method to create sparse differential regulatory network. a Bins used for the analysis are created by taking 100 bp bins on the genome and removing any blacklist regions to form the used bins; b For each TF and factor, the ChIP-Seq signal fold change is averaged over the used bins; c For disease and control cell types K562 and GM12878, respectively, a fold change signal matrix is created with rows or bins and columns of TFs and factors; 4) By using difference of the precision matrix, a differential network is created, which is then made sparse through sampling-based penalization