Polymer Modeling of Whole-Nucleus Diploid Genome Organization

Abstract

The 3D genome organization dictates the genome function in many different aspects. With the development of high-throughput chromosome conformation capture experiments (Hi-C) and super-resolution imaging techniques in recent years, much progress has been made in providing insights into the spatial organization and folding mechanism of chromatin. However, most of the existing methods for 3D structure inference are applied to only a subset of chromosomes under the haploid case and the mechanistic explorations through imaging are mostly qualitative and low-throughput. Here, we propose a data-driven polymer model of the whole-nucleus diploid human genome and use it to study the mechanism of genome organization. Our model not only accurately reproduces both intra- and inter- chromosomal Hi-C contacts, but also quantitatively agrees with imaging results in revealing chromosome territories, the radial distribution of individual chromosomes, and radial positioning of centromeres and A/B compartments. Pair-wise and higher-order patterns of the homolog cross-talk revealed by simulated structural ensemble provide insights into the diploid genome organization in the nuclei. Our model further allows mechanistic explorations of the genome organization from several perspectives. For example, we show that both inter-chromosomal interactions and centromere clustering are essential constraints to the overall genome organization. 1D compartmental component and sequence are also shown to be important factors that affect individual chromosome positioning. In addition, we applied the model to study genome organization in tumors and showed large-scale topological distortions such as compartmental radial positioning. In general, our model provides a powerful tool in reconstructing the whole-nucleus diploid genome organization directly from Hi-C data and reveals various mechanisms in determining genome topology in the nuclei.