Enhancer-promoter interactions are encoded by complex genomic signatures on looping chromatin

By Sean Whalen, Rebecca M. Truty, Katherine S. Pollard. Nature Genetics, 2016

421 citations recorded by the source corpus.

Discriminating the gene target of a distal regulatory element from other nearby transcribed genes is a challenging problem with the potential to illuminate the causal underpinnings of complex diseases. We present TargetFinder, a computational method that reconstructs regulatory landscapes from diverse features along the genome. The resulting models accurately predict individual enhancer–promoter interactions across multiple cell lines with a false discovery rate up to 15 times smaller than that obtained using the closest gene. By evaluating the genomic features driving this accuracy, we uncover interactions between structural proteins, transcription factors, epigenetic modifications, and transcription that together distinguish interacting from non-interacting enhancer–promoter pairs. Most of this signature is not proximal to the enhancers and promoters but instead decorates the looping DNA. We conclude that complex but consistent combinations of marks on the one-dimensional genome encode the three-dimensional structure of fine-scale regulatory interactions.

Subfields: Molecular Biology

View Paper: https://www.semanticscholar.org/paper/b95c401804c13979e517ab553a509d626d01f380

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Molecular Biology

Citations in the corpus

Genome-wide enhancer maps link risk variants to disease genes (2021, background): “We next evaluated other approaches to connect regulatory variants to disease-associated genes, including predictions based on signals from expression quantitative trait loci (eQTLs) 18–21 , three-dimensional contacts 22 , gene set enrichment 23 or other enhancer–gene maps 24–31 (Methods).”
Activity-by-Contact model of enhancer-promoter regulation from thousands of CRISPR perturbations (2019, methodology): “Predictions based on prior machine learning approaches, including correlating chromatin marks with gene expression across cell types, were similarly unsuccessful (see Supplementary Methods) 19,20 .”
Machine Learning and Integrative Analysis of Biomedical Big Data (2019, methodology): “RFE is one of the commonly used wrapper FS algorithms in biomedicine [52,53,58,74] and has been recently applied to integrative analysis [33].”
Predicting enhancer-promoter interaction from genomic sequence with deep neural networks (2016, methodology): “[12], such that positive and negative pairs had similar enhancer-promoter distance distributions.”
Exploiting sequence-based features for predicting enhancer–promoter interactions (2017, methodology): “each positive sample, 20 negative samples were sampled, following the same constraint of distance between the positive pairs (Whalen et al., 2016). Thus, the negative sample size is 20 times of the positive sample size on each cell line. The distance between the paired enhancer and promoter in the d”
GeneHancer: genome-wide integration of enhancers and target genes in GeneCards (2017, background): “More recently, several predictive studies have been carried out to determine enhancer to gene links genome-wide, some utilizing a combination of several approaches (27, 28).”
DNA methylation aging clocks: challenges and recommendations (2019, background): “The identification of the target genes of epigenomic regulatory elements [210], such as enhancers or insulators modified by aging-related changes, may be highly informative and enable subsequent functional exploration in humans.”
UMI-tools: modeling sequencing errors in Unique Molecular Identifiers to improve quantification accuracy (2017, positive): “…was only observed when the rate of DNA polymerase errors was simulated as >0.001, considerably higher than reported error rates for even nonrecombinant Taq DNA polymerase (Rittié and Perbal 2008; Whalen et al. 2016), confirming sequencing errors are likely to be the primary source of UMI errors.”