Fig. 5

HiCNAtra removes sample-dependent and sample-independent biases from the contact maps of cancer cell lines. a-b Interaction heatmaps (500-kb bin) and 1D signals of LNCaP chr 10 (a) and H69AR chr 2 (b) corrected by OneD and HiCNAtra approaches. For the OneD tool, OneD + CN normalization module was used in all the analyses. Copy number track computed by HiCNAtra is shown on top. Each grey dot represents the copy number of a bin. The red line represents the copy number track where any amplitude transition indicates a new CNV region. The second and fourth panels show the OneD-corrected, and HiCNAtra-corrected contact maps, respectively. The third and fifth panels show post-corrected 1D signals (red dots) using OneD and HiCNAtra tools, respectively. The blue line shows the moving average of the 1D signal. c Bean plot of standard deviations per chromosome of 1D signal of raw (orange), OneD-corrected (red) and HiCNAtra-corrected (blue) cis contact maps computed from five cancer datasets (MCF7, LNCaP, PC3, H69AR, and K562). d Bean plot showing Spearman correlations between cis contact frequencies and sample-dependent (CNV) and sample-independent (effective length, GC-content, and mappability) biases in raw (orange), OneD-corrected (red) and HiCNAtra-corrected (blue) contact maps. Correlations are calculated across chromosomes of five cancer datasets