Supplement Materials for Normalization of ChIP - seq data with control Kun

We demonstrate that a proper control sample correlates linearly with the background parts of its corresponding ChIP sample. In the following examples, we first draw the original ChIP vs control bins counts to show the over-abundance of high ChIP count bins due to binding signals. Then we filter the strong binding signals by calling peaks with SPP (Kharchenko et al., 2008) at FDR 0.1 level and exclude all the bins that intersect with the peaks. Then the remaining ChIP and control bin counts show clear linear trend and are roughly symmetric around the normalization factor estimated through NCIS.