Producing genome structure populations with the dynamic and automated PGS software

Abstract

Hi-C technologies are widely used to investigate the spatial organization of genomes. Because genome structures can vary considerably between individual cells of a population, interpreting ensemble-averaged Hi-C data can be challenging, in particular for long-range and inter-chromosomal interactions. We pioneered a probabilistic approach for generating a population of distinct diploid 3D genome structures consistent with all the chromatin-chromatin interaction probabilities from Hi-C experiments. Each structure in the population is a physical model of the genome in 3D. Analysis of these models yields new insights into the causes and the functional properties of the genome’s organization in space and time. We provide a user- friendly software package, called PGS, which runs on local machines (for toy runs) and high- performance computing platforms. PGS takes a genome-wide Hi-C contact frequency matrix along with information about genome segmentation and produces an ensemble of 3D genome structures entirely consistent with the input. The software automatically generates an analysis report, and provides tools to extract and analyze the 3D coordinates of specific domains. Basic Linux command line knowledge is sufficient for using this software. A typical running time of the pipeline is about 3 days with 300 cores on a computer cluster to generate a population of 1,000 diploid genome structures at TAD level resolution.

From Hi-C Contact Map to Three-dimensional Organization of Interphase Human Chromosomes

“…Several data-driven approaches have been developed in order to go from Hi-C to 3D structure of genomes [10][11][12][13][14][15][16][17] (see the summary in [18] for additional related studies). Although these methods are insightful, they do not predict the physical dimensions of the organized chromosomes nor have the methods been validated, especially when the structures are highly heterogeneous.…”

Models of polymer physics for the architecture of the cell nucleus

“…Independent of the underlying physical processes, important computational approaches to reconstruct chromosome threedimensional (3D) conformations have been also developed by optimizing scoring functions of Hi-C data to extract a consensus structure (Duan et al, 2010;Hu et al, 2013;Lesne, Riposo, Roger, Cournac, & Mozziconacci, 2014;Peng et al, 2013;Varoquaux, Ay, Noble, & Vert, 2014;Zhang, Li, Toh, & Sung, 2013), its variability by resampling methods (Baù et al, 2011;Rousseau, Fraser, Ferraiuolo, Dostie, & Blanchette, 2011;Serra et al, 2017), population of individual structures by likelihood maximization of contact data and other information (Giorgietti, Galupa, Nora, et al, 2014;Hua et al, 2018;Kalhor, Tjong, Jayathilaka, Alber, & Chen, 2012;Zhang & Wolynes, 2015). However, they are not discussed here.…”

Revealing Hi-C subcompartments by imputing high-resolution inter-chromosomal chromatin interactions

“…The Hi-C subcompartment predictions from SNIPER can be compared to results based on other analysis approaches and datasets. For example, we expect that the SNIPER predictions of Hi-C subcompartments can be used to further validate and compare with results from polymer simulations (Sanborn et al, 2015;Nuebler et al, 2018), 3D genome structure population modeling Hua et al, 2018), and regulatory communities mining based on whole-genome chromatin interactomes . In addition, recently published new genome-wide mapping methods (Chen et al, 2018;Quinodoz et al, 2018;Beagrie et al, 2017) may provide additional training data other than Hi-C, as well as experimental data validation to improve our method.…”