Noncoding loci without epigenomic signals can be essential for maintaining global chromatin organization and cell viability

Abstract

Noncoding loci without epigenomic signals can be essential for maintaining global chromatin organization and cell viability.

Characterization of network hierarchy reflects cell-state specificity in genome organization

“…Graph embedding is a dimensionality reduction strategy which transforms graph connectivity into distances in the embedding space 52,53 . It is known that the existence of hierarchically nested TADs in Hi-C matrix leads to strong clustering in the contact patterns 80,81 , and at the same time, power-law degree distributions of chromatin contact networks indicate the existence of node heterogeneity 12 . Strong clustering and node heterogeneity characterize a network with intrinsic flow hierarchy 15 .…”

“…Hierarchical networks are well known for their efficient routing guaranteed by the underlining hyperbolic geometry 15 . Chromatin contact network is a complex hierarchical network characterized by the scale-free degree distribution and existence of nested TADs 12 . In a chromatin contact network, where a physical contact can be random or bear the mission of transmitting genetic signals (e.g., enhancer-promoter interactions), long-range interactions serve as bridges and can increase the genetic signal transmission efficiency 15 .…”

“…Flow hierarchy describes an architecture where information flows from higher-grade nodes to lower-grade ones 11 . Chromatin contact networks have a scale-free pattern, which is characterized by the power-law distribution of node degrees 12 . Power-law phenomena in nature imply existence of self-organization patterns and hidden hierarchy 13 .…”

“…With the rapid development of experimental techniques for detecting chromatin structure, computational methods for deciphering chromatin structure are also emerging. However, existing methods for modeling chromatin structure are mainly limited in three-dimensional Euclidean space, which overlooks its functional behavior as a hierarchical network 12,20,21 . Through feature extraction, symbolic data such as texts and networks can be transformed into sets of vectors with representation learning, such a process is called embedding 49-51 .…”

Towards key scientific questions in the diagnosis and treatment of rare diseases: Summary from the 297th Meeting of the Shuangqing Forum

“…Thus far, however, pathogenic variants in genes have only been identified in 20%–40% of these conditions, with most located in protein-coding regions ( NIH, 2022 ; Shire, 2013 ). Non-protein-coding sequences occupy 98% of the human genome and, depending on their localization and specific interactions with DNA, RNA, and proteins, play key regulatory roles in nearly 93% of diseases ( Ding et al, 2021 ; Hindorff et al, 2009 ; Liu et al, 2020 ; Statello et al, 2021 ). Non-Mendelian inheritance patterns (e.g., oligogenic inheritance), gene-environment interactions, epigenetics, and other chemical modifications of biomolecules (e.g., glycosylation) also warrant further attention to understand the pathogenic mechanisms underlying undiagnosed and rare diseases ( Fitz-James & Cavalli, 2022 ; Kousi & Katsanis, 2015 ; Oh & Petronis, 2021 ; Wang et al, 2020 ).…”