Fig. 4From: Chromatin segmentation based on a probabilistic model for read counts explains a large portion of the epigenomeTSS prediction. Chromatin states can be used as binary classifiers for detecting transcription start sites (TSSs). Given a state, the true positives are the bins annotated with that state, with high levels of DNase hypersensitivity and where a known TSS has been annotated. The specificity, or precision, is the number of true positives divided by the number of bins annotated with the state; the sensitivity is the number of true positives divided by the number of bins corresponding to assembly gaps. The state which gives maximum precision has been chosenBack to article page