A novel approach uncovering the biological signal in Hi-C data affected by technical biases

The chromatin interaction assays, particularly Hi-C, enabled detailed studies of chromatin architecture in multiple organisms and model systems, resulting in a deeper understanding of gene expression regulation mediated by epigenetics mechanisms. However, the analysis and interpretation of Hi-C data remain challenging due to technical biases, limiting direct comparisons of datasets obtained in different experiments and laboratories. As a result, removing biases from Hi-C-generated chromatin contact matrices is a critical data analysis step. Our novel approach HiConfidence eliminates biases from the Hi-C data by weighing chromatin contacts according to their consistency between replicates so that low-quality replicates do not influence the result. The algorithm is effective for the analysis of global changes in chromatin structures such as compartments and TADs. We apply the HiConfidence approach to several Hi-C datasets with significant technical biases that could not be analyzed effectively using existing methods, and obtain meaningful biological conclusions. In particular, HiConfidence aids in the study of how changes in the histone acetylation pattern affect chromatin organization in Drosophila cells. Overall design: Drosophila melanogaster Schneider-2 (S2) cells were treated to alter the level of histone acetylation. Depletion of histone deacetylase HDAC1 (HDAC-dep) and inhibition of histone deacetylases with trichostatin A (TSA) (HDAC-inh) increased histone acetylation level. Curcumin treatment inhibiting histone acetyltransferases (HAT-inh) was applied to reduce histone acetylation level. For each condition 2 replicates of treated samples and 2 replicates of control samples were obtained.