Figure 4

Estrogen receptor whole-genome ChIP-chip experiment using xMAN or Affymetrix probe mapping. We applied MAT with xMAN or Affymetrix probe mapping to the estrogen receptor ChIP-chip experiment on Affymetrix human genome 1.0 tiling array set, which consists of 14 arrays covering the non-repetitive human genome at 35 bp resolution.

A) MATscore histogram: The standard deviations estimated from the background NULL distribution are 1.07 and 1.09 using xMAN and Affymetrix probe mapping, respectively. Only the bottom part of the histogram was shown.

B) Scatter plot of false discovery rate (FDR) versus number of true positives. Under each cutoff, the number of true positive is estimated as the number of positive peaks minus the number of negative peaks; the FDR is estimated as the number of negative peaks divided by number of positive peaks. Under the same FDR cutoff, MAT predicts more true positive peaks using xMAN probe mapping than using Affymetrix probe mapping.