Figure 2

Two methods for measuring retention of binding sites. We use three examples to illustrate our two complementary methods and their detection of retention or loss of mouse sites. Both methods start out from mouse binding sites, detected as peaks from the mouse ChIP-seq data [15]. Red lines show the midpoints of mouse sites and the corresponding positions in the human genome. Similarly, the yellow area corresponds to the ends of the mouse site and the corresponding orthologous sequences in human. (A) Retention of a mouse site. The mouse binding region overlaps with a human binding region (peak detected from human data) in the corresponding human orthologous region. Using the binary method, this defines a retained site. The enrichment method compares the ChIP density in the orthologous human region with random regions to calculate fold enrichment. In this case the human ChIP signal is strong which results in a high fold enrichment. (B) Intermediary retention. In this example, a fraction of the binding activity remains in the human orthologous region. This will not be detected using peak finding, and the site will be classified as lost using the binary method. However, the enrichment method will give this region an intermediary fold enrichment score, since there are more ChIP-seq tags than expected by random. (C) Loss of a mouse site. Here, there is no detectable ChIP-seq signal in the orthologous human region. The binary method will classify this site as non-retained and thereby lost in human, while the enrichment will report a fold enrichment near zero, since the enrichment is similar to that of random regions.