Multi-scale structural community organisation of the human genome

Abstract

BackgroundStructural interaction frequency matrices between all genome loci are now experimentally achievable thanks to high-throughput chromosome conformation capture technologies. This ensues a new methodological challenge for computational biology which consists in objectively extracting from these data the structural motifs characteristic of genome organisation.ResultsWe deployed the fast multi-scale community mining algorithm based on spectral graph wavelets to characterise the networks of intra-chromosomal interactions in human cell lines. We observed that there exist structural domains of all sizes up to chromosome length and demonstrated that the set of structural communities forms a hierarchy of chromosome segments. Hence, at all scales, chromosome folding predominantly involves interactions between neighbouring sites rather than the formation of links between distant loci.ConclusionsMulti-scale structural decomposition of human chromosomes provides an original framework to question structural organisation and its relationship to functional regulation across the scales. By construction the proposed methodology is independent of the precise assembly of the reference genome and is thus directly applicable to genomes whose assembly is not fully determined.Electronic supplementary materialThe online version of this article (doi:10.1186/s12859-017-1616-x) contains supplementary material, which is available to authorized users.

SpectralTAD: an R package for defining a hierarchy of topologically associated domains using spectral clustering

“…The matrix C is square and symmetric around the diagonal representing self-interacting regions. Our method relies on the fact that the 3D chromosome can be thought of as a naturally occurring graph [16,51]. Traditionally, a graph G(V, E) is represented by a series of nodes V connected by edges E. These graphs are summarized in an adjacency matrix A ij , where entry ij indicates the number of edges between node i and node j.…”

“…To date, most methods were developed to find these single meta-TADs instead of focusing on the hierarchy of the TAD structures [ 13 – 15 ]. While interesting insights can be gleaned from the meta-TADs [ 16 ], work has shown that smaller sub-TADs are specifically associated with gene regulation [ 10 , 17 , 18 ]. For example, it has been found that genes associated with limb malformation in rats are specifically controlled through interactions within sub-TADs [ 18 ].…”

Exploring chromatin hierarchical organization via Markov State Modelling

“…Several previous studies using a hierarchical clustering approach for the analysis of Hi-C data are based on the a priori assumption of the existence of structural hierarchy in chromatin [38–40]. While the work by Boulos et al is free from such an assumption [37], it employs a tunable scale parameter in establishing the hierarchy. Our approach is based on an energy landscape representation of the chromatin interaction network, which is formalized and explored via a Markov State Model.…”

“…In data-driven studies, on the other hand, experimental Hi-C interaction maps are used for extracting information on statistically significant chromatin interactions, for defining topologically associating domains (TADs) and A/B compartments [35, 36], and for the 3D reconstruction of chromatin. Several algorithms have been introduced to study the hierarchical organization of chromatin and its correlation with the distribution of various epigenetic features [37–40], including graph-based approaches for exploring sparse networks of Hi-C interaction peaks, as well as ChIA-PET and HiChIP interaction pairs [41–43]. A recent work by Pancaldi et al .…”

“…To date, many methods developed for domain detection [23, 45–47] essentially adopt an image segmentation approach aimed at identifying domain regions as a function of short-range interactions along the chromosome, and domain boundary positions are often highly sensitive to the choice of heuristic tuning parameters [48]. Recent network-based methods incorporate effects of long-range interactions in characterizing structural organization [37, 39, 49] and observe spatial couplings at multiple scales associated with the regulation of gene expression [50]. Spatial reconstructions of chromatin using Hi-C interaction data yield consensus structures [51, 52] or ensembles of possible chromosomal conformations [53, 54], providing an overall picture of chromatin organization [55].…”