Figure 2

Classification of epigenetic marks in human CD4+ cells. (A) The frequency of sequence tags in a typical ChIP-SEQ experiment (H3K4me3) in 4 kb regions [36] and a simulated null distribution (red line). Vertical lines represent a false discovery rate (FDR) of 1% from each side in the null distribution (Depleted and Enriched), leaving the rest as (Neutral). (B) Based on the criteria in (A), the number of selected enriched (dark gray), depleted (white), or neutral (medium gray) regions for each histone mark. (C) The SVM classification of three representative methylated histone marks. Error bars represent standard deviations (s.d.) of 10 SVM predictions. (D) Receiver Operating Characteristic (ROC) curves for the classification of three representative methylated histone marks (solid lines) and the corresponding curves for classification with randomized sample labels (dotted lines).