MATCHA: Probing multi-way chromatin interaction with hypergraph representation learning

Machine and Deep Learning Methods for Predicting 3D Genome Organization

Three-dimensional (3D) chromatin interactions, such as enhancer-promoter interactions (EPIs), loops, topologically associating domains (TADs), and A/B compartments, play critical roles in a wide range of cellular processes by regulating gene expression. Recent development of chromatin conformation capture technologies has enabled genome-wide profiling of various 3D structures, even with single cells. However, current catalogs of 3D structures remain incomplete and unreliable due to differences in technology, tools, and low data resolution. Machine learning methods have emerged as an alternative to obtain missing 3D interactions and/or improve resolution. Such methods frequently use genome annotation data (ChIP-seq, DNAse-seq, etc.), DNA sequencing information (k-mers and transcription factor binding site (TFBS) motifs), and other genomic properties to learn the associations between genomic features and chromatin interactions. In this review, we discuss computational tools for predicting three types of 3D interactions (EPIs, chromatin interactions, and TAD boundaries) and analyze their pros and cons. We also point out obstacles to the computational prediction of 3D interactions and suggest future research directions.

Current and future directions in network biology

Summary

Network biology is an interdisciplinary field bridging computational and biological sciences that has proved pivotal in advancing the understanding of cellular functions and diseases across biological systems and scales. Although the field has been around for two decades, it remains nascent. It has witnessed rapid evolution, accompanied by emerging challenges. These stem from various factors, notably the growing complexity and volume of data together with the increased diversity of data types describing different tiers of biological organization. We discuss prevailing research directions in network biology, focusing on molecular/cellular networks but also on other biological network types such as biomedical knowledge graphs, patient similarity networks, brain networks, and social/contact networks relevant to disease spread. In more detail, we highlight areas of inference and comparison of biological networks, multimodal data integration and heterogeneous networks, higher-order network analysis, machine learning on networks, and network-based personalized medicine. Following the overview of recent breakthroughs across these five areas, we offer a perspective on future directions of network biology. Additionally, we discuss scientific communities, educational initiatives, and the importance of fostering diversity within the field. This article establishes a roadmap for an immediate and long-term vision for network biology.

Availability and implementation

Not applicable.

Statistical modeling and significance estimation of multi-way chromatin contacts with HyperloopFinder

Recent advances in chromatin conformation capture technologies, such as SPRITE and Pore-C, have enabled the detection of simultaneous contacts among multiple chromatin loci. This has made it possible to investigate the cooperative transcriptional regulation involving multiple genes and regulatory elements at the resolution of a single molecule. However, these technologies are unavoidably subject to the random polymer looping effect and technical biases, making it challenging to distinguish genuine regulatory relationships directly from random polymer interactions. Here, we present HyperloopFinder, a method for identifying regulatory multi-way chromatin contacts (hyperloops) by jointly modeling the random polymer looping effect and technical biases to estimate the statistical significance of multi-way contacts. The results show that our model can accurately estimate the expected interaction frequency of multi-way contacts based on the distance distribution of pairwise contacts, revealing that most multi-way contacts can be formed by randomly linking the pairwise contacts adjacent to each other. Moreover, we observed the spatial colocalization of the interaction sites of hyperloops from image-based data. Our results also revealed that hyperloops can function as scaffolds for the cooperation among multiple genes and regulatory elements. In summary, our work contributes novel insights into higher-order chromatin structures and functions and has the potential to enhance our understanding of transcriptional regulation and other cellular processes.

Machine and deep learning methods for predicting 3D genome organization

Three-Dimensional (3D) chromatin interactions, such as enhancer-promoter interactions (EPIs), loops, Topologically Associating Domains (TADs), and A/B compartments play critical roles in a wide range of cellular processes by regulating gene expression. Recent development of chromatin conformation capture technologies has enabled genome-wide profiling of various 3D structures, even with single cells. However, current catalogs of 3D structures remain incomplete and unreliable due to differences in technology, tools, and low data resolution. Machine learning methods have emerged as an alternative to obtain missing 3D interactions and/or improve resolution. Such methods frequently use genome annotation data (ChIP-seq, DNAse-seq, etc.), DNA sequencing information (k-mers, Transcription Factor Binding Site (TFBS) motifs), and other genomic properties to learn the associations between genomic features and chromatin interactions. In this review, we discuss computational tools for predicting three types of 3D interactions (EPIs, chromatin interactions, TAD boundaries) and analyze their pros and cons. We also point out obstacles of computational prediction of 3D interactions and suggest future research directions.

Computational methods for analysing multiscale 3D genome organization

Recent progress in whole-genome mapping and imaging technologies has enabled the characterization of the spatial organization and folding of the genome in the nucleus. In parallel, advanced computational methods have been developed to leverage these mapping data to reveal multiscale three-dimensional (3D) genome features and to provide a more complete view of genome structure and its connections to genome functions such as transcription. Here, we discuss how recently developed computational tools, including machine-learning-based methods and integrative structure-modelling frameworks, have led to a systematic, multiscale delineation of the connections among different scales of 3D genome organization, genomic and epigenomic features, functional nuclear components and genome function. However, approaches that more comprehensively integrate a wide variety of genomic and imaging datasets are still needed to uncover the functional role of 3D genome structure in defining cellular phenotypes in health and disease.

A Lightweight Framework For Chromatin Loop Detection at the Single-Cell Level

Single-cell Hi-C (scHi-C) has made it possible to analyze chromatin organization at the single-cell level. However, scHi-C experiments generate inherently sparse data, which poses a challenge for loop calling methods. The existing approach performs significance tests across the imputed dense contact maps, leading to substantial computational overhead and loss of information at the single-cell level. To overcome this limitation, a lightweight framework called scGSLoop is proposed, which sets a new paradigm for scHi-C loop calling by adapting the training and inferencing strategies of graph-based deep learning to leverage the sequence features and 1D positional information of genomic loci. With this framework, sparsity is no longer a challenge, but rather an advantage that the model leverages to achieve unprecedented computational efficiency. Compared to existing methods, scGSLoop makes more accurate predictions and is able to identify more loops that have the potential to play regulatory roles in genome functioning. Moreover, scGSLoop preserves single-cell information by identifying a distinct group of loops for each individual cell, which not only enables an understanding of the variability of chromatin looping states between cells, but also allows scGSLoop to be extended for the investigation of multi-connected hubs and their underlying mechanisms.

PhyGCN: Pre-trained Hypergraph Convolutional Neural Networks with Self-supervised Learning

Hypergraphs are powerful tools for modeling complex interactions across various domains, including biomedicine. However, learning meaningful node representations from hypergraphs remains a challenge. Existing supervised methods often lack generalizability, thereby limiting their real-world applications. We propose a new method, Pre-trained Hypergraph Convolutional Neural Networks with Self-supervised Learning (PhyGCN), which leverages hypergraph structure for self-supervision to enhance node representations. PhyGCN introduces a unique training strategy that integrates variable hyperedge sizes with self-supervised learning, enabling improved generalization to unseen data. Applications on multi-way chromatin interactions and polypharmacy side-effects demonstrate the effectiveness of PhyGCN. As a generic framework for high-order interaction datasets with abundant unlabeled data, PhyGCN holds strong potential for enhancing hypergraph node representations across various domains.

Mega-Enhancer Bodies Organize Neuronal Long Genes in the Cerebellum

Dynamic regulation of gene expression plays a key role in establishing the diverse neuronal cell types in the brain. Recent findings in genome biology suggest that three-dimensional (3D) genome organization has important, but mechanistically poorly understood functions in gene transcription. Beyond local genomic interactions between promoters and enhancers, we find that cerebellar granule neurons undergoing differentiation in vivo exhibit striking increases in long-distance genomic interactions between transcriptionally active genomic loci, which are separated by tens of megabases within a chromosome or located on different chromosomes. Among these interactions, we identify a nuclear subcompartment enriched for near-megabase long enhancers and their associated neuronal long genes encoding synaptic or signaling proteins. Neuronal long genes are differentially recruited to this enhancer-dense subcompartment to help shape the transcriptional identities of granule neuron subtypes in the cerebellum. SPRITE analyses of higher-order genomic interactions, together with IGM-based 3D genome modeling and imaging approaches, reveal that the enhancer-dense subcompartment forms prominent nuclear structures, which we term mega-enhancer bodies. These novel nuclear bodies reside in the nuclear periphery, away from other transcriptionally active structures, including nuclear speckles located in the nuclear interior. Together, our findings define additional layers of higher-order 3D genome organization closely linked to neuronal maturation and identity in the brain.

Prediction of gene co-expression from chromatin contacts with graph attention network

MOTIVATION

The technology of high-throughput chromatin conformation capture (Hi-C) allows genome-wide measurement of chromatin interactions. Several studies have shown statistically significant relationships between gene-gene spatial contacts and their co-expression. It is desirable to uncover epigenetic mechanisms of transcriptional regulation behind such relationships using computational modeling. Existing methods for predicting gene co-expression from Hi-C data use manual feature engineering or unsupervised learning, which either limits the prediction accuracy or lacks interpretability.

RESULTS

To address these issues, we propose HiCoEx (Hi-C predicts gene co-expression), a novel end-to-end framework for explainable prediction of gene co-expression from Hi-C data based on graph neural network. We apply graph attention mechanism to a gene contact network inferred from Hi-C data to distinguish the importance among different neighboring genes of each gene, and learn the gene representation to predict co-expression in a supervised and task-specific manner. Then, from the trained model, we extract the learned gene embeddings as a model interpretation to distill biological insights. Experimental results show that HiCoEx can learn gene representation from 3D genomics signals automatically to improve prediction accuracy, and make the black box model explainable by capturing some biologically meaningful patterns, e.g., in a gene contact network, the common neighbors of two central genes might contribute to the co-expression of the two central genes through sharing enhancers.

AVAILABILITY AND IMPLEMENTATION

The source code is freely available at https://github.com/JieZheng-ShanghaiTech/HiCoEx.

SUPPLEMENTARY INFORMATION

Supplementary data are available at Bioinformatics online.

Deciphering multi-way interactions in the human genome

Chromatin architecture, a key regulator of gene expression, can be inferred using chromatin contact data from chromosome conformation capture, or Hi-C. However, classical Hi-C does not preserve multi-way contacts. Here we use long sequencing reads to map genome-wide multi-way contacts and investigate higher order chromatin organization in the human genome. We use hypergraph theory for data representation and analysis, and quantify higher order structures in neonatal fibroblasts, biopsied adult fibroblasts, and B lymphocytes. By integrating multi-way contacts with chromatin accessibility, gene expression, and transcription factor binding, we introduce a data-driven method to identify cell type-specific transcription clusters. We provide transcription factor-mediated functional building blocks for cell identity that serve as a global signature for cell types.

Mapping higher order chromatin architecture is important. Here the authors use long sequencing reads to map genome-wide multi-way contacts and investigate higher order chromatin organisation; they use hypergraph theory for data representation and analysis, and apply this to different cell types.

Machine Learning Methods for Exploring Sequence Determinants of 3D Genome Organization

In higher eukaryotic cells, chromosomes are folded inside the nucleus. Recent advances in whole-genome mapping technologies have revealed the multiscale features of 3D genome organization that are intertwined with fundamental genome functions. However, DNA sequence determinants that modulate the formation of 3D genome organization remain poorly characterized. In the past few years, predicting 3D genome organization based on DNA sequence features has become an active area of research. Here, we review the recent progress in computational approaches to unraveling important sequence elements for 3D genome organization. In particular, we discuss the rapid development of machine learning-based methods that facilitate the connections between DNA sequence features and 3D genome architectures at different scales. While much progress has been made in developing predictive models for revealing important sequence features for 3D genome organization, new research is urgently needed to incorporate multi-omic data and enhance model interpretability, further advancing our understanding of gene regulation mechanisms through the lens of 3D genome organization.

Chromatin Hubs: A biological and computational outlook

This review discusses our current understanding of chromatin biology and bioinformatics under the unifying concept of “chromatin hubs.” The first part reviews the biology of chromatin hubs, including chromatin–chromatin interaction hubs, chromatin hubs at the nuclear periphery, hubs around macromolecules such as RNA polymerase or lncRNAs, and hubs around nuclear bodies such as the nucleolus or nuclear speckles. The second part reviews existing computational methods, including enhancer–promoter interaction prediction, network analysis, chromatin domain callers, transcription factory predictors, and multi-way interaction analysis. We introduce an integrated model that makes sense of the existing evidence. Understanding chromatin hubs may allow us (i) to explain long-unsolved biological questions such as interaction specificity and redundancy of mechanisms, (ii) to develop more realistic kinetic and functional predictions, and (iii) to explain the etiology of genomic disease.

Learning representations of chromatin contacts using a recurrent neural network identifies genomic drivers of conformation

Despite the availability of chromatin conformation capture experiments, discerning the relationship between the 1D genome and 3D conformation remains a challenge, which limits our understanding of their affect on gene expression and disease. We propose Hi-C-LSTM, a method that produces low-dimensional latent representations that summarize intra-chromosomal Hi-C contacts via a recurrent long short-term memory neural network model. We find that these representations contain all the information needed to recreate the observed Hi-C matrix with high accuracy, outperforming existing methods. These representations enable the identification of a variety of conformation-defining genomic elements, including nuclear compartments and conformation-related transcription factors. They furthermore enable in-silico perturbation experiments that measure the influence of cis-regulatory elements on conformation.

3D chromatin architecture and transcription regulation in cancer

Chromatin has distinct three-dimensional (3D) architectures important in key biological processes, such as cell cycle, replication, differentiation, and transcription regulation. In turn, aberrant 3D structures play a vital role in developing abnormalities and diseases such as cancer. This review discusses key 3D chromatin structures (topologically associating domain, lamina-associated domain, and enhancer-promoter interactions) and corresponding structural protein elements mediating 3D chromatin interactions [CCCTC-binding factor, polycomb group protein, cohesin, and Brother of the Regulator of Imprinted Sites (BORIS) protein] with a highlight of their associations with cancer. We also summarise the recent development of technologies and bioinformatics approaches to study the 3D chromatin interactions in gene expression regulation, including crosslinking and proximity ligation methods in the bulk cell population (ChIA-PET and HiChIP) or single-molecule resolution (ChIA-drop), and methods other than proximity ligation, such as GAM, SPRITE, and super-resolution microscopy techniques.

The 3D Genome Structure of Single Cells

The spatial organization of the genome in the cell nucleus is pivotal to cell function. However, how the 3D genome organization and its dynamics influence cellular phenotypes remains poorly understood. The very recent development of single-cell technologies for probing the 3D genome, especially single-cell Hi-C (scHi-C), has ushered in a new era of unveiling cell-to-cell variability of 3D genome features at an unprecedented resolution. Here, we review recent developments in computational approaches to the analysis of scHi-C, including data processing, dimensionality reduction, imputation for enhancing data quality, and the revealing of 3D genome features at single-cell resolution. While much progress has been made in computational method development to analyze single-cell 3D genomes, substantial future work is needed to improve data interpretation and multimodal data integration, which are critical to reveal fundamental connections between genome structure and function among heterogeneous cell populations in various biological contexts.