Figure 3

ROC-like curve for ENCODE spike-in data using either xMAN or Affymetrix probe mapping. We applied MAT with either xMAN or Affymetrix probe mapping to the spike-in data. xMAN and Affymetrix mapping achieved 100% and 96% True Positive Rate (TPR) at 0% False Discovery Rate (FDR) cutoff, respectively. A MAT prediction is considered correct if the center of the predicted region lies in the actual spike-in fragment. Please note that False Discovery Rate instead of False Positive Rate is used in this ROC-like curve.