Fig. 6

Local DNA sequence composition deviations associated with cognate and noncognate CC-seq sites. a Scatterplots of human TOP2 CC-seq hits on the Watson strand vs the Crick strand at each 3 bp offset site, in: observed data; a dataset sampled from a simulation where all 3 bp offset sites had identical Watson and Crick values (‘100% cognate’); randomised observed data where Crick hits were shuffled. Noncognates (green) and cognates (gold) were defined as those 3 bp offset sites with significantly different, or not significantly different Watson and Crick hits, respectively (two-sample, two-tailed Poisson test; p ≤ 0.05). Sites with fewer than eight hits in total (Watson + Crick) were excluded due to insufficient statistical power. b Percentage of Top2 CC-seq sites that were classed cognate vs noncognate in: observed, simulated 100% cognate, and randomised as defined in (a) for S. cerevisiae and human VP16-treated samples. S. cerevisiae cognate simulation and randomisation use pdr1∆ mre11∆ dataset. c, d Cognate (gold) and noncognate (green) Top2 CC-seq signal from VP16-treated pdr1∆ S. cerevisiae cells (c) or RPE-1 cells (d), within example regions. The data were smoothed according to local density (c), or binned at 10 Kbp resolution and smoothed with a sliding 10-bin Hanning window (d). e Average nucleotide composition over a 32 bp window centred on the highly cognate (p > 0.95) or (inferred) noncognate (p ≤ 0.05) Top2 dyad axis, in VP16-treated pdr1∆ mre11∆ S. cerevisiae and human RPE-1 cells. Values reported are for the top strand only. f Heatmaps of nucleotide composition deviations from the local average, for noncognate (top) and highly cognate (middle) sites; and the difference between the two (bottom). For the upper two panels, the absolute log₂ deviations of the four nucleotides were summed at each position, and are expressed on a colour-scale from yellow to red. For the bottom panel, the absolute log₂ fold difference between highly cognate and noncognate were summed at each position and are expressed on a colour-scale from white to black. HpM Hits per million mapped reads per base pair