Fig. 1

Improving the DNAse Hi-C protocol allows the generation of high-quality Hi-C maps. a Quality metrics of Hi-C datasets. Data are grouped according to the protocol employed for library construction. Each dot represents an independent Hi-C library preparation. The dataset names are explained in Table 1, and all the details of each protocol are described in the “Methods” section. Note that Ramani et al. ([21]) performed Hi-C on mouse samples, whereas other data were from human cells, which could explain some of the differences between “Ramani et al. (reanalysed)” and other samples. The reported pairs percentage indicates the mapping efficiency; cis-interactions reflect noise levels; FR-excess indicates the overrepresentation of reads in the forward–reverse (inward) orientation, a signature of undigested or unligated DNA (DEnds); when possible, estimated DEnds were corrected using information about biotinylated linker incorporation (see “Methods” section) and are shown on the corrected DEnds estimation plot. The significance of differences between groups was estimated using the Mann–Whitney test. b Representative Hi-C data obtained using different protocols for human K562 cells and for mouse brain cells (in the case of reanalysed Ramani et al. data). For K562 cells, each Hi-C heatmap shows a comparison between results obtained using the protocol from [23] (above the diagonal line) and data obtained using the biotin fill-in protocol developed in this study (below the diagonal line). All data were downsampled to the same sequencing depth, and the number in the top right corner indicates the values selected on the Juicebox colour slider. Additional heatmaps showing representative genomic regions are shown in Additional file 1: Fig. AS1