Inferring Radial Organization of Chromosomal Territories from HI-C Data

Abstract

Listen

Translate article

In the mammalian genome, the chromosomes inside the nucleus are organized radially into discrete bodies, known as chromosomal territories (CTs). The radial organization of CTs plays a vital role in the nuclear organization and disruption of this organization appears to contribute to abnormalities in the higher-order chromatin structure of the genome in disease. With the help of microscopy imaging techniques, it has been already shown that the radial organization of the interphase CTs follows cell-type-specific non-random distribution. Though the imaging techniques can produce visually compelling accurate results of radial organization of CTs, their application has been limited. On the other hand, data generated by chromosome conformation capture techniques (e.g. 3C, 4C, 5C, and Hi-C) which can reveal the snapshot of the spatial genome organization at a particular time point, are becoming widely available. Therefore, techniques to infer the radial organization of CTs from this data would be greatly advantageous. To infer the radial organization of CTs from the Hi-C contact information, here, we propose a computational method consisting of a force-directed network layout algorithm and a regression method. This method generates multiple 3D network configurations from the bulk Hi-C data and applies a regression method that uses chromosome gene density and length based information to get the averaged distance of the CTs from the nucleus center based on all of the generated structures. The CT distance data predicted by this method showed a high correlation with microscopy imaging data for the lymphoblastoid, skin fibroblast and breast epithelium cells. We further explore the ability of this method to detect meaningful differences in CT position in disease states or biochemical perturbations.