Human Enhancers Are Fragile and Prone to Deactivating Mutations

Chromatin accessibility: biological functions, molecular mechanisms and therapeutic application

The dynamic regulation of chromatin accessibility is one of the prominent characteristics of eukaryotic genome. The inaccessible regions are mainly located in heterochromatin, which is multilevel compressed and access restricted. The remaining accessible loci are generally located in the euchromatin, which have less nucleosome occupancy and higher regulatory activity. The opening of chromatin is the most important prerequisite for DNA transcription, replication, and damage repair, which is regulated by genetic, epigenetic, environmental, and other factors, playing a vital role in multiple biological progresses. Currently, based on the susceptibility difference of occupied or free DNA to enzymatic cleavage, solubility, methylation, and transposition, there are many methods to detect chromatin accessibility both in bulk and single-cell level. Through combining with high-throughput sequencing, the genome-wide chromatin accessibility landscape of many tissues and cells types also have been constructed. The chromatin accessibility feature is distinct in different tissues and biological states. Research on the regulation network of chromatin accessibility is crucial for uncovering the secret of various biological processes. In this review, we comprehensively introduced the major functions and mechanisms of chromatin accessibility variation in different physiological and pathological processes, meanwhile, the targeted therapies based on chromatin dynamics regulation are also summarized.

ChromDL: a next-generation regulatory DNA classifier

MOTIVATION

Predicting the regulatory function of non-coding DNA using only the DNA sequence continues to be a major challenge in genomics. With the advent of improved optimization algorithms, faster GPU speeds, and more intricate machine-learning libraries, hybrid convolutional and recurrent neural network architectures can be constructed and applied to extract crucial information from non-coding DNA.

RESULTS

Using a comparative analysis of the performance of thousands of Deep Learning architectures, we developed ChromDL, a neural network architecture combining bidirectional gated recurrent units, convolutional neural networks, and bidirectional long short-term memory units, which significantly improves upon a range of prediction metrics compared to its predecessors in transcription factor binding site, histone modification, and DNase-I hyper-sensitive site detection. Combined with a secondary model, it can be utilized for accurate classification of gene regulatory elements. The model can also detect weak transcription factor binding as compared to previously developed methods and has the potential to help delineate transcription factor binding motif specificities.

AVAILABILITY AND IMPLEMENTATION

The ChromDL source code can be found at https://github.com/chrishil1/ChromDL.

A survey on algorithms to characterize transcription factor binding sites

Transcription factors (TFs) are key regulatory proteins that control the transcriptional rate of cells by binding short DNA sequences called transcription factor binding sites (TFBS) or motifs. Identifying and characterizing TFBS is fundamental to understanding the regulatory mechanisms governing the transcriptional state of cells. During the last decades, several experimental methods have been developed to recover DNA sequences containing TFBS. In parallel, computational methods have been proposed to discover and identify TFBS motifs based on these DNA sequences. This is one of the most widely investigated problems in bioinformatics and is referred to as the motif discovery problem. In this manuscript, we review classical and novel experimental and computational methods developed to discover and characterize TFBS motifs in DNA sequences, highlighting their advantages and drawbacks. We also discuss open challenges and future perspectives that could fill the remaining gaps in the field.

ChromDL: A Next-Generation Regulatory DNA Classifier

Motivation

Predicting the regulatory function of non-coding DNA using only the DNA sequence continues to be a major challenge in genomics. With the advent of improved optimization algorithms, faster GPU speeds, and more intricate machine learning libraries, hybrid convolutional and recurrent neural network architectures can be constructed and applied to extract crucial information from non-coding DNA.

Results

Using a comparative analysis of the performance of thousands of Deep Learning (DL) architectures, we developed ChromDL, a neural network architecture combining bidirectional gated recurrent units (BiGRU), convolutional neural networks (CNNs), and bidirectional long short-term memory units (BiLSTM), which significantly improves upon a range of prediction metrics compared to its predecessors in transcription factor binding site (TFBS), histone modification (HM), and DNase-I hypersensitive site (DHS) detection. Combined with a secondary model, it can be utilized for accurate classification of gene regulatory elements. The model can also detect weak transcription factor (TF) binding with higher accuracy as compared to previously developed methods and has the potential to accurately delineate TF binding motif specificities.

Availability

The ChromDL source code can be found at https://github.com/chrishil1/ChromDL .

GRAFIMO: Variant and haplotype aware motif scanning on pangenome graphs

Transcription factors (TFs) are proteins that promote or reduce the expression of genes by binding short genomic DNA sequences known as transcription factor binding sites (TFBS). While several tools have been developed to scan for potential occurrences of TFBS in linear DNA sequences or reference genomes, no tool exists to find them in pangenome variation graphs (VGs). VGs are sequence-labelled graphs that can efficiently encode collections of genomes and their variants in a single, compact data structure. Because VGs can losslessly compress large pangenomes, TFBS scanning in VGs can efficiently capture how genomic variation affects the potential binding landscape of TFs in a population of individuals. Here we present GRAFIMO (GRAph-based Finding of Individual Motif Occurrences), a command-line tool for the scanning of known TF DNA motifs represented as Position Weight Matrices (PWMs) in VGs. GRAFIMO extends the standard PWM scanning procedure by considering variations and alternative haplotypes encoded in a VG. Using GRAFIMO on a VG based on individuals from the 1000 Genomes project we recover several potential binding sites that are enhanced, weakened or missed when scanning only the reference genome, and which could constitute individual-specific binding events. GRAFIMO is available as an open-source tool, under the MIT license, at https://github.com/pinellolab/GRAFIMO and https://github.com/InfOmics/GRAFIMO.

Transcription factors (TFs) are key regulatory proteins and mutations occurring in their binding sites can alter the normal transcriptional landscape of a cell and lead to disease states. Pangenome variation graphs (VGs) efficiently encode genomes from a population of individuals and their genetic variations. GRAFIMO is an open-source tool that extends the traditional PWM scanning procedure to VGs. By scanning for potential TBFS in VGs, GRAFIMO can simultaneously search thousands of genomes while accounting for SNPs, indels, and structural variants. GRAFIMO reports motif occurrences, their statistical significance, frequency, and location within the reference or alternative haplotypes in a given VG. GRAFIMO makes it possible to study how genetic variation affects the binding landscape of known TFs within a population of individuals.

Learning and interpreting the gene regulatory grammar in a deep learning framework

Deep neural networks (DNNs) have achieved state-of-the-art performance in identifying gene regulatory sequences, but they have provided limited insight into the biology of regulatory elements due to the difficulty of interpreting the complex features they learn. Several models of how combinatorial binding of transcription factors, i.e. the regulatory grammar, drives enhancer activity have been proposed, ranging from the flexible TF billboard model to the stringent enhanceosome model. However, there is limited knowledge of the prevalence of these (or other) sequence architectures across enhancers. Here we perform several hypothesis-driven analyses to explore the ability of DNNs to learn the regulatory grammar of enhancers. We created synthetic datasets based on existing hypotheses about combinatorial transcription factor binding site (TFBS) patterns, including homotypic clusters, heterotypic clusters, and enhanceosomes, from real TF binding motifs from diverse TF families. We then trained deep residual neural networks (ResNets) to model the sequences under a range of scenarios that reflect real-world multi-label regulatory sequence prediction tasks. We developed a gradient-based unsupervised clustering method to extract the patterns learned by the ResNet models. We demonstrated that simulated regulatory grammars are best learned in the penultimate layer of the ResNets, and the proposed method can accurately retrieve the regulatory grammar even when there is heterogeneity in the enhancer categories and a large fraction of TFBS outside of the regulatory grammar. However, we also identify common scenarios where ResNets fail to learn simulated regulatory grammars. Finally, we applied the proposed method to mouse developmental enhancers and were able to identify the components of a known heterotypic TF cluster. Our results provide a framework for interpreting the regulatory rules learned by ResNets, and they demonstrate that the ability and efficiency of ResNets in learning the regulatory grammar depends on the nature of the prediction task.

Stable enhancers are active in development, and fragile enhancers are associated with evolutionary adaptation

Background

Despite continual progress in the identification and characterization of trait- and disease-associated variants that disrupt transcription factor (TF)-DNA binding, little is known about the distribution of TF binding deactivating mutations (deMs) in enhancer sequences. Here, we focus on elucidating the mechanism underlying the different densities of deMs in human enhancers.

Results

We identify two classes of enhancers based on the density of nucleotides prone to deMs. Firstly, fragile enhancers with abundant deM nucleotides are associated with the immune system and regular cellular maintenance. Secondly, stable enhancers with only a few deM nucleotides are associated with the development and regulation of TFs and are evolutionarily conserved. These two classes of enhancers feature different regulatory programs: the binding sites of pioneer TFs of FOX family are specifically enriched in stable enhancers, while tissue-specific TFs are enriched in fragile enhancers. Moreover, stable enhancers are more tolerant of deMs due to their dominant employment of homotypic TF binding site (TFBS) clusters, as opposed to the larger-extent usage of heterotypic TFBS clusters in fragile enhancers. Notably, the sequence environment and chromatin context of the cognate motif, other than the motif itself, contribute more to the susceptibility to deMs of TF binding.

Conclusions

This dichotomy of enhancer activity is conserved across different tissues, has a specific footprint in epigenetic profiles, and argues for a bimodal evolution of gene regulatory programs in vertebrates. Specifically encoded stable enhancers are evolutionarily conserved and associated with development, while differently encoded fragile enhancers are associated with the adaptation of species.

Electronic supplementary material

The online version of this article (10.1186/s13059-019-1750-z) contains supplementary material, which is available to authorized users.

Dichotomy in redundant enhancers points to presence of initiators of gene regulation

Background

The regulatory landscape of a gene locus often consists of several functionally redundant enhancers establishing phenotypic robustness and evolutionary stability of its regulatory program. However, it is unclear what mechanisms are employed by redundant enhancers to cooperatively orchestrate gene expression.

Results

By comparing redundant enhancers to single enhancers (enhancers present in a single copy in a gene locus), we observed that the DNA sequence encryption differs between these two classes of enhancers, suggesting a difference in their regulatory mechanisms. Initiator enhancers, which are a subset of redundant enhancers and show similar sequence encryption to single enhancers, differ from the rest of redundant enhancers in their sequence encryption, evolutionary conservation and proximity to target genes. Genes hosting initiator enhancers in their loci feature elevated levels of expression. Initiator enhancers show a high level of 3D chromatin contacts with both transcription start sites and regular enhancers, suggesting their roles as primary activators and intermediate catalysts of gene expression, through which the regulatory signals of redundant enhancers are propagated to the target genes. In addition, GWAS and eQTLs variants are significantly enriched in initiator enhancers compared to redundant enhancers, suggesting a key functional role these sequences play in gene regulation.

Conclusions

The specific characteristics and widespread abundance of initiator enhancers advocate for a possible universal hierarchical mechanism of tissue-specific gene regulation involving multiple redundant enhancers acting through initiator enhancers.

Electronic supplementary material

The online version of this article (10.1186/s12864-018-5335-0) contains supplementary material, which is available to authorized users.

Prediction of gene regulatory enhancers across species reveals evolutionarily conserved sequence properties

Genomic regions with gene regulatory enhancer activity turnover rapidly across mammals. In contrast, gene expression patterns and transcription factor binding preferences are largely conserved between mammalian species. Based on this conservation, we hypothesized that enhancers active in different mammals would exhibit conserved sequence patterns in spite of their different genomic locations. To investigate this hypothesis, we evaluated the extent to which sequence patterns that are predictive of enhancers in one species are predictive of enhancers in other mammalian species by training and testing two types of machine learning models. We trained support vector machine (SVM) and convolutional neural network (CNN) classifiers to distinguish enhancers defined by histone marks from the genomic background based on DNA sequence patterns in human, macaque, mouse, dog, cow, and opossum. The classifiers accurately identified many adult liver, developing limb, and developing brain enhancers, and the CNNs outperformed the SVMs. Furthermore, classifiers trained in one species and tested in another performed nearly as well as classifiers trained and tested on the same species. We observed similar cross-species conservation when applying the models to human and mouse enhancers validated in transgenic assays. This indicates that many short sequence patterns predictive of enhancers are largely conserved. The sequence patterns most predictive of enhancers in each species matched the binding motifs for a common set of TFs enriched for expression in relevant tissues, supporting the biological relevance of the learned features. Thus, despite the rapid change of active enhancer locations between mammals, cross-species enhancer prediction is often possible. Our results suggest that short sequence patterns encoding enhancer activity have been maintained across more than 180 million years of mammalian evolution.

Finding the Sources of Missing Heritability within Rare Variants Through Simulation

Thousands of genome-wide association studies (GWAS) have been conducted to identify the genetic variants associated with complex disorders. However, only a small proportion of phenotypic variances can be explained by the reported variants. Moreover, many GWAS failed to identify genetic variants associated with disorders displaying hereditary features. The “missing heritability” problem can be partly explained by rare variants. We simulated a causality scenario that gestational ages, a quantitative trait that can distinguish preterm (<37 weeks) and term births, were significantly correlated with the rare variant aggregations at 1000 single-nucleotide polymorphism loci. These 1000 simulated causal rare variants were embedded into randomly selected subsets of 9642 promoter regions from the 1000 Genomes Project genotypic data according to different proportions of causal rare variants within the embedded promoters. Through analysis of the correlations between rare variant aggregations and gestational ages, we found that the embedded promoters as a whole showed weaker genetic association when the proportion of causal rare variants decreased, and no individual embedded promoters showed genetic association when the proportion of causal rare variants was smaller than 0.4. Our analyses indicate that association signals can be greatly diluted when causal rare variants are dispersedly and sparsely distributed in the genome, accounting for an important source of missing heritability.

SNPDelScore: combining multiple methods to score deleterious effects of noncoding mutations in the human genome

SUMMARY

Addressing deleterious effects of noncoding mutations is an essential step towards the identification of disease-causal mutations of gene regulatory elements. Several methods for quantifying the deleteriousness of noncoding mutations using artificial intelligence, deep learning and other approaches have been recently proposed. Although the majority of the proposed methods have demonstrated excellent accuracy on different test sets, there is rarely a consensus. In addition, advanced statistical and artificial learning approaches used by these methods make it difficult porting these methods outside of the labs that have developed them. To address these challenges and to transform the methodological advances in predicting deleterious noncoding mutations into a practical resource available for the broader functional genomics and population genetics communities, we developed SNPDelScore, which uses a panel of proposed methods for quantifying deleterious effects of noncoding mutations to precompute and compare the deleteriousness scores of all common SNPs in the human genome in 44 cell lines. The panel of deleteriousness scores of a SNP computed using different methods is supplemented by functional information from the GWAS Catalog, libraries of transcription factor-binding sites, and genic characteristics of mutations. SNPDelScore comes with a genome browser capable of displaying and comparing large sets of SNPs in a genomic locus and rapidly identifying consensus SNPs with the highest deleteriousness scores making those prime candidates for phenotype-causal polymorphisms.

AVAILABILITY AND IMPLEMENTATION

https://www.ncbi.nlm.nih.gov/research/snpdelscore/.

SUPPLEMENTARY INFORMATION

Supplementary data are available at Bioinformatics online.

Quantifying deleterious effects of regulatory variants

The majority of genome-wide association study (GWAS) risk variants reside in non-coding DNA sequences. Understanding how these sequence modifications lead to transcriptional alterations and cell-to-cell variability can help unraveling genotype-phenotype relationships. Here, we describe a computational method, dubbed CAPE, which calculates the likelihood of a genetic variant deactivating enhancers by disrupting the binding of transcription factors (TFs) in a given cellular context. CAPE learns sequence signatures associated with putative enhancers originating from large-scale sequencing experiments (such as ChIP-seq or DNase-seq) and models the change in enhancer signature upon a single nucleotide substitution. CAPE accurately identifies causative cis-regulatory variation including expression quantitative trait loci (eQTLs) and DNase I sensitivity quantitative trait loci (dsQTLs) in a tissue-specific manner with precision superior to several currently available methods. The presented method can be trained on any tissue-specific dataset of enhancers and known functional variants and applied to prioritize disease-associated variants in the corresponding tissue.

iRegNet3D: three-dimensional integrated regulatory network for the genomic analysis of coding and non-coding disease mutations

The mechanistic details of most disease-causing mutations remain poorly explored within the context of regulatory networks. We present a high-resolution three-dimensional integrated regulatory network (iRegNet3D) in the form of a web tool, where we resolve the interfaces of all known transcription factor (TF)-TF, TF-DNA and chromatin-chromatin interactions for the analysis of both coding and non-coding disease-associated mutations to obtain mechanistic insights into their functional impact. Using iRegNet3D, we find that disease-associated mutations may perturb the regulatory network through diverse mechanisms including chromatin looping. iRegNet3D promises to be an indispensable tool in large-scale sequencing and disease association studies.

Emergence of the Noncoding Cancer Genome: A Target of Genetic and Epigenetic Alterations

The emergence of whole-genome annotation approaches is paving the way for the comprehensive annotation of the human genome across diverse cell and tissue types exposed to various environmental conditions. This has already unmasked the positions of thousands of functional cis-regulatory elements integral to transcriptional regulation, such as enhancers, promoters, and anchors of chromatin interactions that populate the noncoding genome. Recent studies have shown that cis-regulatory elements are commonly the targets of genetic and epigenetic alterations associated with aberrant gene expression in cancer. Here, we review these findings to showcase the contribution of the noncoding genome and its alteration in the development and progression of cancer. We also highlight the opportunities to translate the biological characterization of genetic and epigenetic alterations in the noncoding cancer genome into novel approaches to treat or monitor disease.

SIGNIFICANCE

The majority of genetic and epigenetic alterations accumulate in the noncoding genome throughout oncogenesis. Discriminating driver from passenger events is a challenge that holds great promise to improve our understanding of the etiology of different cancer types. Advancing our understanding of the noncoding cancer genome may thus identify new therapeutic opportunities and accelerate our capacity to find improved biomarkers to monitor various stages of cancer development. Cancer Discov; 6(11); 1215-29. ©2016 AACR.