Fig. 6
From: In silico prediction of high-resolution Hi-C interaction matrices
Assessing HiC-Reg predicted counts in new test cell lines. a Distance-stratified Pearson’s correlation plot using models trained on one cell line (columns) and tested on a different cell line (rows) for chromosome 17 considering all five cell lines: Huvec, Hmec, Nhek, Gm12878, K562. Each plot (except the ones on the diagonal) shows the distance-stratified correlation curves when training using the same cell line (blue), a different cell line using MULTI-CELL (red) and WINDOW (cyan) features, using the ensemble of all predictions from a different cell line (orange), and when simply transferring counts (green). The plots on the diagonal show the CV performance. b Area under the curve (AUC) for distance-stratified correlation in chromosome 17. Shown are the AUCs for the different training cell line models for both feature types (red and green), ensembles from both feature types, as well as transfer count when predicting in a new cell line. Source data are provided as a Source Data file.
