Skip to main content
. 2022 Jun 15;8(24):eabn9215. doi: 10.1126/sciadv.abn9215

Fig. 2. EagleC outperforms existing methods in both precision and recall.

Fig. 2.

(A) (Left) Genome-wide Hi-C map and the predicted SVs (black circles) in BT-474. (Right) Enlarged Hi-C maps of the indicated SV regions. The SV breakpoints detected by different methods are highlighted using different marks on each map. (B) Number of SVs predicted by EagleC and Hi-C breakfinder in BT-474, HCC1954, and MCF7 cells with levels of validation by orthogonal methods. For Hi-C breakfinder, we only counted SVs reported at the 10-kb resolution. (C) Precision and recall rates of SVs that can only be predicted by EagleC, SVs that can only be predicted by Hi-C breakfinder, and SVs that can be predicted by both methods. (D) The number of SVs predicted by EagleC and Hi-C breakfinder in additional 26 cancer samples with both Hi-C and WGS data available. (E and F) Recall rates and precision rates of SVs predicted by EagleC and Hi-C breakfinder in 26 cancer samples. Each dot represents an individual sample. The P values were computed using the two-sided Wilcoxon signed-rank test. (G) Size distributions of intrachromosomal SVs detected by EagleC and Hi-C breakfinder. Data were merged from 29 cancer samples. (H) Number of different range of SVs predicted by EagleC and Hi-C breakfinder with validation ratios by WGS. Short SVs, intrachromosomal SVs with breakpoint distance less than 1 Mb; long SVs, intrachromosomal SVs with breakpoint distance greater than 1 Mb; TL, interchromosomal translocations.