Table 2.

ΔBIC from the intercept-only model for four measures of biological plausibility of contacts

Model	Non-activated	Activated
a: Promoter: match to Cao et al.
MPPC	−338.0	−245.8
CHiCAGO	−332.2	−331.4
MPPC + CHiCAGO	a−411.6	a−358.8
b: Promoter: link to active chromatin
MPPC	−1134.2	−812.3
CHiCAGO	−659.5	−560.1
MPPC + CHiCAGO	a−1231.1	a−924.0
c: Validation: overlap baited promoter
MPPC	−404.8	−347.9
CHiCAGO	−430.0	−419.2
MPPC + CHiCAGO	a−541.7	a−499.0
d: Validation: expression at linked promoter
MPPC	−1571.0	a−1329.2
CHiCAGO	−871.9	−430.2
MPPC + CHiCAGO	a−1640.3	−1318.7

The best fitting model (lowest ΔBIC) is highlighted by ^a. a–d are defined in full in the “Methods” section. Briefly,

a whether the bait-prey pair corresponds to published CD4 ⁺ T cell promoter-enhancer networks [17];

b whether the prey fragment overlaps active chromatin states defined by [7];

c whether the prey overlaps a gene promoter;

d the level of expression of a gene associated with the prey fragment. In all cases, a robust clustered model was used to account for repeated observations at the prey fragment