Chromatin interaction analysis with paired-end tag sequencing (ChIA-PET) is a relatively new technique for studying interactions between genomic loci involving a protein of interest. It aims to provide some insights into the molecular mechanisms of transcriptional regulation by identifying interactions between genes and distal regulatory elements such as enhancers. Unlike the other "C"-based methods for studying chromatin organization, the ChIA-PET procedure produces an explicit measure of non-specific events during the proximity ligation step. However, most existing bioinformatics analyses only use this information to make general statements on the quality of each library.
Here, we propose a novel approach using edgeR to identify significant interactions based on differences between counts for specific and non-specific ligation products. This improves the relevance of the detected interactions by accounting for variability in non-specific ligation across the genome. This is particularly important for cancer studies where the observed ligation patterns can be affected by copy number variation or structural rearrangements. Replicates can also be explicitly incorporated into the statistical model to account for biological variability. We demonstrate the applicability of this approach with publicly available ChIA-PET data on several immortalized cell lines.