Fig. 4
From: 7C: Computational Chromosome Conformation Capture by Correlation of ChIP-seq at CTCF motifs
Prediction performance in HeLa cells using 7C trained in GM12878 cells. a ROC curve of prediction performance of six selected TF ChIP-seq data sets. The 7C model was trained using ChIP-seq and true loop data in human GM1287 but loops were predicted using ChIP-seq data of the same TFs in HeLa cells and true loop data in HeLa cells. b Precision-Recall curves for the same analysis as in (a). c Prediction performance as auPRC (top) and auROC (bottom) in HeLa for the six TF ChIP-seq data sets (x-axis) and 7C models trained for the specific TF (left), 7C with parameters averaged across all 124 TF models (center), and 7C with parameters as average of the 10 best performing TF ChIP-seq data sets (right). d Example region on human chromosome 21 with genes, RAD21 ChIP-seq data in HeLa, CTCF motifs, true loops in HeLa cells according to Hi-C and ChIA-PET (green arcs) and predicted chromatin loops from RAD21 ChIP-seq data in HeLa (blue arcs)