Testing whether disruption of chromatin accessibility preferentially occurs at the same regulatory elements in two cell types using conditional P-value distributions; illustration of the basis of our approach. (A) Hypothetical example of two cell types showing disruption of chromatin accessibility primarily at the same regulatory elements when comparing mutant and wild-type mice in each cell type. There are also some regulatory elements differentially accessible between the mutant and wild type in only one of the two cell types (such as peak 4), but the overlap between the two cell types exceeds what is expected by chance. The resulting distribution of P-values in cell type A, when restricting to regulatory elements that are differentially accessible in cell type B, shows an overrepresentation of P-values close to zero. This overrepresentation is not observed when restricting to regulatory elements that are not differentially accessible in cell type B. (B) Like A, but with different regulatory elements disrupted in cell types. There are also some regulatory elements differentially accessible between the mutant and wild type in both cell types (such as peak 4), but the overlap between the two cell types does not exceed what is expected by chance. The resulting distribution of P-values in cell type A shows the same overrepresentation close to zero regardless of whether we restrict to regulatory elements that are differential in cell type B or regulatory elements that are not differential in cell type B.
