Abstract:Cis-regulatory elements (CREs) contact target genes to regulate their transcription, but CRE interactions are complex and predicting their function is challenging. Contemporary approaches to predict CRE targets generally assume one-to-one associations and linear genomic proximity. These methods fail to account for long-range CRE contacts caused by folding of the genome, for circumstances in which a single CRE regulates multiple genes, or, conversely, when a single gene is regulated by multiple CREs. To address these challenges and investigate three-dimensional (3D) CRE communities, we developed BOUQUET, an integrative, graph-theory-based approach that uses multiple aspects of genome topology to probe communities of CREs, their bound apparatus, and their target genes. Using this approach, we uncover communities of CREs that are undetectable by existing unbiased methods, such as those with interactions that span insulating loop boundaries, and regulatory communities that comprise clustered promoter elements. A subset of these communities, which we term “3D super-enhancer CRE communities” (3DSE CRE communities), accrues exceptional amounts of regulatory apparatus at cell identity-defining genes and evolves the super-enhancer concept. In imaging assays, we observe co-localization and co-expression of in-community genes within puncta of transcriptional co-factors, consistent with microcompartment biology. Genome-wide, pairs of genes within the same 3D CRE community are co-expressed in single cells, and pairs of CREs are co-accessible, suggesting both chromatin activity and transcription within a topologically defined community is coordinated. Our model more accurately reflects the complexity of CRE interaction networks, underscores the need for approaches that incorporate spatial organization in the study of enhancer biology, and provides a framework for the reinterpretation of SEs, their condensates, and their target genes through the lens of 3D chromatin structure.

CREATE: cell-type-specific cis-regulatory elements identification via discrete embedding

Identification, Design, and Application of Noncoding Cis-Regulatory Elements

Single-cell Multi-Scale Footprinting Reveals the Modular Organization of DNA Regulatory Elements

DIRECT-NET: an Efficient Method to Discover Cis-Regulatory Elements and Construct Regulatory Networks from Single-Cell Multiomics Data

A Tiling-Deletion-Based Genetic Screen for Cis-Regulatory Element Identification in Mammalian Cells

DeepCAPE: A Deep Convolutional Neural Network for the Accurate Prediction of Enhancers.

An attention-based deep neural network model to detect cis-regulatory elements at the single-cell level from multi-omics data

CREdb: A comprehensive database of Cis-Regulatory Elements and their activity in human cells and tissues

DeepEnhancer: Predicting Enhancers by Convolutional Neural Networks.

Discrete Latent Embedding of Single-Cell Chromatin Accessibility Sequencing Data for Uncovering Cell Heterogeneity

EPInformer: A Scalable Deep Learning Framework for Gene Expression Prediction by Integrating Promoter-enhancer Sequences with Multimodal Epigenomic Data

Topology-informed regulatory element collections coordinate cell identity gene expression programs

Inferring predictive genetic models and regulatory elements by deep learning of cross-species single-cell gene expression landscapes

Predicting Enhancers with Deep Convolutional Neural Networks

Identifying and Classifying Enhancers by Dinucleotide-Based Auto-Cross Covariance and Attention-Based Bi-LSTM

Interpreting -Regulatory Interactions from Large-Scale Deep Neural Networks for Genomics

Machine Learning Uncovers Cell Identity Regulator by Histone Code.

DeepRegFinder: deep learning-based regulatory elements finder

Probabilistic association of differentially expressed genes with cis -regulatory elements

Enhancer-MDLF: a novel deep learning framework for identifying cell-specific enhancers

Multiplex profiling of developmental cis- regulatory elements with quantitative single-cell expression reporters