Skip to main content
Fig. 2 | Genome Biology

Fig. 2

From: Modeling double strand break susceptibility to interrogate structural variation in cancer

Fig. 2

Accurate models of DSB frequency built from chromatin and sequence features. ad Random forest regression model predictions built upon 11 genomic features at 50 kb resolution compared to observed DSB frequencies for four datasets: NHEK DSBCapture, NHEK BLESS, K562 BLISS, and MCF7 BLISS. The y values reflect the sequencing depth of each dataset. The models’ predictions are all highly correlated with the observed data, as shown by the noted Pearson’s correlations (p < 2.2e−16 for each dataset). eh The predictive features ranked by variable importance, a measure of how useful a particular feature is for the model (see the “Methods” section)

Back to article page