Figure 11

Performance of RSSPA. A pseudo ROC method is employed here for assessment of the performance of RSSPA versus binary segmentation. The x-axis corresponds to the FPR(1 – specificity) and the y-axis corresponds to sensitivity. The solid curves are representative of RNA pol II data derived from chromosome 1 (ENCODE array) and the dotted curve is representative of RNA pol II data derived from all chromosomes sampled by the ENCODE array. The solid curves exhibit the performance of RSSPA (orange) versus binary segmentation using a pScore of 50 (blue), 40 (red) and 30 (magenta). The dotted curve represents the performance of RSSPA across the entire ENCODE region. The positive regions in the pseudo ROC are derived from the known 5'ends (RefSeq) ± 500 bp of the first exon(and UTR). The negative regions are derived from intergenic regions as well as the inner-most introns of transcripts. While the true performances of the algorithms are not precisely determined in the pseudo ROC curves the estimation of their relative performance is accurate.