Fig. 1

Schematic example of analysis in a single CHi-C region. a raw counts derived from the CHi-C experiment and the decay profile around the bait estimated by negative binomial (NB) regression. b residuals after NB regression represent the signal after adjusting for expected exponential decay around the bait. These residuals can be compared against a null distribution to generate p-values, in the same way CHiCAGO scores are generated – separately for each fragment, independent of local patterns in the signal. c Peaky extends the inference by considering the joint distributions of residuals across the region. This example shows three proposed direct contacts which could be jointly responsible for the spatial distribution of NB residuals, with solid lines indicating the individual decay functions fit in our joint model and the dashed line their predicted joint effect. The position of the bait fragment is indicated by the dotted line and chromosome position is shown in kb relative to the bait. Peaky inference at each prey fragment is based on the marginal posterior probability of a contact (MPPC), defined as the proportion of MCMC samples in which that fragment was selected as a direct contact with a positive peak height