A neural network based model effectively predicts enhancers from clinical ATAC-seq samples

Predmoter-cross-species prediction of plant promoter and enhancer regions

Motivation

Identifying cis-regulatory elements (CREs) is crucial for analyzing gene regulatory networks. Next generation sequencing methods were developed to identify CREs but represent a considerable expenditure for targeted analysis of few genomic loci. Thus, predicting the outputs of these methods would significantly cut costs and time investment.

Results

We present Predmoter, a deep neural network that predicts base-wise Assay for Transposase Accessible Chromatin using sequencing (ATAC-seq) and histone Chromatin immunoprecipitation DNA-sequencing (ChIP-seq) read coverage for plant genomes. Predmoter uses only the DNA sequence as input. We trained our final model on 21 species for 13 of which ATAC-seq data and for 17 of which ChIP-seq data was publicly available. We evaluated our models on Arabidopsis thaliana and Oryza sativa. Our best models showed accurate predictions in peak position and pattern for ATAC- and histone ChIP-seq. Annotating putatively accessible chromatin regions provides valuable input for the identification of CREs. In conjunction with other in silico data, this can significantly reduce the search space for experimentally verifiable DNA-protein interaction pairs.

Availability and implementation

The source code for Predmoter is available at: https://github.com/weberlab-hhu/Predmoter. Predmoter takes a fasta file as input and outputs h5, and optionally bigWig and bedGraph files.

Epigenomic landscape of the human dorsal root ganglion: sex differences and transcriptional regulation of nociceptive genes

Gene expression is influenced by chromatin architecture via controlled access of regulatory factors to DNA. To better understand regulation of gene expression in the human dorsal root ganglion (hDRG) we used bulk and spatial transposase-accessible chromatin technology followed by sequencing (ATAC-seq). We detected a total of 3005 differentially accessible chromatin regions (DARs) between sexes using bulk ATAC-seq. DARs in female hDRG mapped mainly to the X chromosome. In males, DARs were found in autosomal genes. We also found differential transcription factor binding motifs within DARs. EGR1/3 and SP1/4 were abundant in females, and JUN, FOS and other AP-1 family members in males. With the aim of dissecting the open chromatin profile in hDRG neurons, we used spatial ATAC-seq. Consistent with our bulk ATAC-seq data, most of the DARs in female hDRG were located in X chromosome genes. Neuron cluster showed higher chromatin accessibility in GABAergic, glutamatergic, and interferon-related genes in females, and in Ca2+-signaling-related genes in males. Sex differences in open chromatin transcription factor binding sites in neuron-proximal barcodes were consistent with the bulk data, having EGR1 transcription factor activity in females and AP-1 family members in males. Accordingly, we showed higher expression of EGR1 in female hDRG compared to male with in-situ hybridization. Our findings point to epigenomic sex differences in the hDRG that likely underlie divergent transcriptional responses that determine mechanistic sex differences in pain.

Cis-regulatory atlas of primary human CD4+ T cells

Cis-regulatory elements (CRE) are critical for coordinating gene expression programs that dictate cell-specific differentiation and homeostasis. Recently developed self-transcribing active regulatory region sequencing (STARR-Seq) has allowed for genome-wide annotation of functional CREs. Despite this, STARR-Seq assays are only employed in cell lines, in part, due to difficulties in delivering reporter constructs. Herein, we implemented and validated a STARR-Seq-based screen in human CD4+ T cells using a non-integrating lentiviral transduction system. Lenti-STARR-Seq is the first example of a genome-wide assay of CRE function in human primary cells, identifying thousands of functional enhancers and negative regulatory elements (NREs) in human CD4+ T cells. We find an unexpected difference in nucleosome organization between enhancers and NRE: enhancers are located between nucleosomes, whereas NRE are occupied by nucleosomes in their endogenous locations. We also describe chromatin modification, eRNA production, and transcription factor binding at both enhancers and NREs. Our findings support the idea of silencer repurposing as enhancers in alternate cell types. Collectively, these data suggest that Lenti-STARR-Seq is a successful approach for CRE screening in primary human cell types, and provides an atlas of functional CREs in human CD4+ T cells.

Enhancer-silencer transitions in the human genome

Dual-function regulatory elements (REs), acting as enhancers in some cellular contexts and as silencers in others, have been reported to facilitate the precise gene regulatory response to developmental signals in Drosophila melanogaster. However, with few isolated examples detected, dual-function REs in mammals have yet to be systematically studied. We herein investigated this class of REs in the human genome and profiled their activity across multiple cell types. Focusing on enhancer–silencer transitions specific to the development of T cells, we built an accurate deep learning classifier of REs and identified about 12,000 silencers active in primary peripheral blood T cells that act as enhancers in embryonic stem cells. Compared with regular silencers, these dual-function REs are evolving under stronger purifying selection and are enriched for mutations associated with disease phenotypes and altered gene expression. In addition, they are enriched in the loci of transcriptional regulators, such as transcription factors (TFs) and chromatin remodeling genes. Dual-function REs consist of two intertwined but largely distinct sets of binding sites bound by either activating or repressing TFs, depending on the type of RE function in a given cell line. This indicates the recruitment of different TFs for different regulatory modes and a complex DNA sequence composition of these REs with dual activating and repressive encoding. With an estimated >6% of cell type–specific human silencers acting as dual-function REs, this overlooked class of REs requires a specific investigation on how their inherent functional plasticity might be a contributing factor to human diseases.

CoRE-ATAC: A deep learning model for the functional classification of regulatory elements from single cell and bulk ATAC-seq data

Cis-Regulatory elements (cis-REs) include promoters, enhancers, and insulators that regulate gene expression programs via binding of transcription factors. ATAC-seq technology effectively identifies active cis-REs in a given cell type (including from single cells) by mapping accessible chromatin at base-pair resolution. However, these maps are not immediately useful for inferring specific functions of cis-REs. For this purpose, we developed a deep learning framework (CoRE-ATAC) with novel data encoders that integrate DNA sequence (reference or personal genotypes) with ATAC-seq cut sites and read pileups. CoRE-ATAC was trained on 4 cell types (n = 6 samples/replicates) and accurately predicted known cis-RE functions from 7 cell types (n = 40 samples) that were not used in model training (mean average precision = 0.80, mean F1 score = 0.70). CoRE-ATAC enhancer predictions from 19 human islet samples coincided with genetically modulated gain/loss of enhancer activity, which was confirmed by massively parallel reporter assays (MPRAs). Finally, CoRE-ATAC effectively inferred cis-RE function from aggregate single nucleus ATAC-seq (snATAC) data from human blood-derived immune cells that overlapped with known functional annotations in sorted immune cells, which established the efficacy of these models to study cis-RE functions of rare cells without the need for cell sorting. ATAC-seq maps from primary human cells reveal individual- and cell-specific variation in cis-RE activity. CoRE-ATAC increases the functional resolution of these maps, a critical step for studying regulatory disruptions behind diseases.

Machine learning meets omics: applications and perspectives

The innovation of biotechnologies has allowed the accumulation of omics data at an alarming rate, thus introducing the era of 'big data'. Extracting inherent valuable knowledge from various omics data remains a daunting problem in bioinformatics. Better solutions often need some kind of more innovative methods for efficient handlings and effective results. Recent advancements in integrated analysis and computational modeling of multi-omics data helped address such needs in an increasingly harmonious manner. The development and application of machine learning have largely advanced our insights into biology and biomedicine and greatly promoted the development of therapeutic strategies, especially for precision medicine. Here, we propose a comprehensive survey and discussion on what happened, is happening and will happen when machine learning meets omics. Specifically, we describe how artificial intelligence can be applied to omics studies and review recent advancements at the interface between machine learning and the ever-widest range of omics including genomics, transcriptomics, proteomics, metabolomics, radiomics, as well as those at the single-cell resolution. We also discuss and provide a synthesis of ideas, new insights, current challenges and perspectives of machine learning in omics.

Employing core regulatory circuits to define cell identity

The interplay between extrinsic signaling and downstream gene networks controls the establishment of cell identity during development and its maintenance in adult life. Advances in next-generation sequencing and single-cell technologies have revealed additional layers of complexity in cell identity. Here, we review our current understanding of transcription factor (TF) networks as key determinants of cell identity. We discuss the concept of the core regulatory circuit as a set of TFs and interacting factors that together define the gene expression profile of the cell. We propose the core regulatory circuit as a comprehensive conceptual framework for defining cellular identity and discuss its connections to cell function in different contexts.

Epigenomic landscape of human colorectal cancer unveils an aberrant core of pan-cancer enhancers orchestrated by YAP/TAZ

Cancer is characterized by pervasive epigenetic alterations with enhancer dysfunction orchestrating the aberrant cancer transcriptional programs and transcriptional dependencies. Here, we epigenetically characterize human colorectal cancer (CRC) using de novo chromatin state discovery on a library of different patient-derived organoids. By exploring this resource, we unveil a tumor-specific deregulated enhancerome that is cancer cell-intrinsic and independent of interpatient heterogeneity. We show that the transcriptional coactivators YAP/TAZ act as key regulators of the conserved CRC gained enhancers. The same YAP/TAZ-bound enhancers display active chromatin profiles across diverse human tumors, highlighting a pan-cancer epigenetic rewiring which at single-cell level distinguishes malignant from normal cell populations. YAP/TAZ inhibition in established tumor organoids causes extensive cell death unveiling their essential role in tumor maintenance. This work indicates a common layer of YAP/TAZ-fueled enhancer reprogramming that is key for the cancer cell state and can be exploited for the development of improved therapeutic avenues.

The role of epigenetic deregulation in colorectal cancer (CRC) is not fully understood yet. Here the authors use patient-derived organoids, epigenomics and single-cell RNA-seq to reveal that YAP/TAZ are key regulators that bind to active enhancers in CRC and promote tumour survival.

Fish-Ing for Enhancers in the Heart

Precise control of gene expression is crucial to ensure proper development and biological functioning of an organism. Enhancers are non-coding DNA elements which play an essential role in regulating gene expression. They contain specific sequence motifs serving as binding sites for transcription factors which interact with the basal transcription machinery at their target genes. Heart development is regulated by intricate gene regulatory network ensuring precise spatiotemporal gene expression program. Mutations affecting enhancers have been shown to result in devastating forms of congenital heart defect. Therefore, identifying enhancers implicated in heart biology and understanding their mechanism is key to improve diagnosis and therapeutic options. Despite their crucial role, enhancers are poorly studied, mainly due to a lack of reliable way to identify them and determine their function. Nevertheless, recent technological advances have allowed rapid progress in enhancer discovery. Model organisms such as the zebrafish have contributed significant insights into the genetics of heart development through enabling functional analyses of genes and their regulatory elements in vivo. Here, we summarize the current state of knowledge on heart enhancers gained through studies in model organisms, discuss various approaches to discover and study their function, and finally suggest methods that could further advance research in this field.

Self-organizing maps with variable neighborhoods facilitate learning of chromatin accessibility signal shapes associated with regulatory elements

BACKGROUND

Assigning chromatin states genome-wide (e.g. promoters, enhancers, etc.) is commonly performed to improve functional interpretation of these states. However, computational methods to assign chromatin state suffer from the following drawbacks: they typically require data from multiple assays, which may not be practically feasible to obtain, and they depend on peak calling algorithms, which require careful parameterization and often exclude the majority of the genome. To address these drawbacks, we propose a novel learning technique built upon the Self-Organizing Map (SOM), Self-Organizing Map with Variable Neighborhoods (SOM-VN), to learn a set of representative shapes from a single, genome-wide, chromatin accessibility dataset to associate with a chromatin state assignment in which a particular RE is prevalent. These shapes can then be used to assign chromatin state using our workflow.

RESULTS

We validate the performance of the SOM-VN workflow on 14 different samples of varying quality, namely one assay each of A549 and GM12878 cell lines and two each of H1 and HeLa cell lines, primary B-cells, and brain, heart, and stomach tissue. We show that SOM-VN learns shapes that are (1) non-random, (2) associated with known chromatin states, (3) generalizable across sets of chromosomes, and (4) associated with magnitude and multimodality. We compare the accuracy of SOM-VN chromatin states against the Clustering Aggregation Tool (CAGT), an unsupervised method that learns chromatin accessibility signal shapes but does not associate these shapes with REs, and we show that overall precision and recall is increased when learning shapes using SOM-VN as compared to CAGT. We further compare enhancer state assignments from SOM-VN in signals above a set threshold to enhancer state assignments from Predicting Enhancers from ATAC-seq Data (PEAS), a deep learning method that assigns enhancer chromatin states to peaks. We show that the precision-recall area under the curve for the assignment of enhancer states is comparable to PEAS.

CONCLUSIONS

Our work shows that the SOM-VN workflow can learn relationships between REs and chromatin accessibility signal shape, which is an important step toward the goal of assigning and comparing enhancer state across multiple experiments and phenotypic states.

A pitfall for machine learning methods aiming to predict across cell types

Machine learning models that predict genomic activity are most useful when they make accurate predictions across cell types. Here, we show that when the training and test sets contain the same genomic loci, the resulting model may falsely appear to perform well by effectively memorizing the average activity associated with each locus across the training cell types. We demonstrate this phenomenon in the context of predicting gene expression and chromatin domain boundaries, and we suggest methods to diagnose and avoid the pitfall. We anticipate that, as more data becomes available, future projects will increasingly risk suffering from this issue.

Analyzing a putative enhancer of optic disc morphology

Background

Genome-wide association studies have identified the CDC7-TGFBR3 intergenic region on chromosome 1 to be strongly associated with optic disc area size. The mechanism of its function remained unclear until new data on eQTL markers emerged from the Genotype-Tissue Expression project. The target region was found to contain a strong silencer of the distal (800 kb) Transcription Factor (TF) gene GFI1 (Growth Factor Independent Transcription Repressor 1) specifically in neuroendocrine cells (pituitary gland). GFI1 has also been reported to be involved in the development of sensory neurons and hematopoiesis. Therefore, GFI1, being a developmental gene, is likely to affect optic disc area size by altering the expression of the associated genes via long-range interactions.

Results

Distribution of haplotypes in the putative enhancer region has been assessed using the data on four continental supergroups generated by the 1000 Genomes Project. The East Asian (EAS) populations were shown to manifest a highly homogenous unimodal haplotype distribution pattern within the region with the major haplotype occurring with the frequency of 0.9. Another European specific haplotype was observed with the frequency of 0.21. The major haplotype appears to be involved in silencing GFI1repressor gene expression, which might be the cause of increased optic disc area characteristic of the EAS populations.

The enhancer/eQTL region overlaps AluJo element, which implies that this particular regulatory element is primate-specific and confined to few tissues.

Conclusion

Population specific distribution of GFI1 enhancer alleles may predispose certain ethnic groups to glaucoma.

Combining signal and sequence to detect RNA polymerase initiation in ATAC-seq data

The assay for transposase-accessible chromatin followed by sequencing (ATAC-seq) is an inexpensive protocol for measuring open chromatin regions. ATAC-seq is also relatively simple and requires fewer cells than many other high-throughput sequencing protocols. Therefore, it is tractable in numerous settings where other high throughput assays are challenging to impossible. Hence it is important to understand the limits of what can be inferred from ATAC-seq data. In this work, we leverage ATAC-seq to predict the presence of nascent transcription. Nascent transcription assays are the current gold standard for identifying regions of active transcription, including markers for functional transcription factor (TF) binding. We combine mapped short reads from ATAC-seq with the underlying peak sequence, to determine regions of active transcription genome-wide. We show that a hybrid signal/sequence representation classified using recurrent neural networks (RNNs) can identify these regions across different cell types.

Exploration of a diversity of computational and statistical measures of association for genome-wide genetic studies

Background

The principal line of investigation in Genome Wide Association Studies (GWAS) is the identification of main effects, that is individual Single Nucleotide Polymorphisms (SNPs) which are associated with the trait of interest, independent of other factors. A variety of methods have been proposed to this end, mostly statistical in nature and differing in assumptions and type of model employed. Moreover, for a given model, there may be multiple choices for the SNP genotype encoding. As an alternative to statistical methods, machine learning methods are often applicable. Typically, for a given GWAS, a single approach is selected and utilized to identify potential SNPs of interest. Even when multiple GWAS are combined through meta-analyses within a consortium, each GWAS is typically analyzed with a single approach and the resulting summary statistics are then utilized in meta-analyses.

Results

In this work we use as case studies a Type 2 Diabetes (T2D) and a breast cancer GWAS to explore a diversity of applicable approaches spanning different methods and encoding choices. We assess similarity of these approaches based on the derived ranked lists of SNPs and, for each GWAS, we identify a subset of representative approaches that we use as an ensemble to derive a union list of top SNPs. Among these are SNPs which are identified by multiple approaches as well as several SNPs identified by only one or a few of the less frequently used approaches. The latter include SNPs from established loci and SNPs which have other supporting lines of evidence in terms of their potential relevance to the traits.

Conclusions

Not every main effect analysis method is suitable for every GWAS, but for each GWAS there are typically multiple applicable methods and encoding options. We suggest a workflow for a single GWAS, extensible to multiple GWAS from consortia, where representative approaches are selected among a pool of suitable options, to yield a more comprehensive set of SNPs, potentially including SNPs that would typically be missed with the most popular analyses, but that could provide additional valuable insights for follow-up.

Electronic supplementary material

The online version of this article (10.1186/s13040-019-0201-4) contains supplementary material, which is available to authorized users.