Taken together, CTCF binding data for multiple individuals show that mutations can be buffered to maintain the levels of binding signal, particularly at highly conserved sites, and this effect cannot be explained solely by the flexibility of CTCF's sequence consensus. We asked whether mechanisms potentially accountable for such buffering would also affect the relationship between sequence and binding in the absence of mutations. Training an interaction linear model across the whole set of mapped CTCF binding sites revealed that conservation consistently weakens the relationship between PWM score and the binding intensity (P = 1.9e-7; Figure 5e). Thus, CTCF binding to evolutionarily conserved sites may generally have a reduced dependence on sequence.
