Biology subjects

Hughes, J. R.

Publications and source records attributed to Hughes, J. R..

At least 19 recordsLinked to original sources

bioRxiv2026

Learning quality scores for chromatin accessibility bigWig tracks using Machine Learning

High-throughput chromatin accessibility assays such as bulk and single-cell ATAC-seq have generated large collections of processed signal tracks in bigWig format, which are widely used for visualisation, data integration, and Machine Learning (ML)-based analyses. Despite their central role, systematic quality control (QC) frameworks operating directly at the level of bigWig signal tracks remain underdeveloped. This gap limits the ability to assess data reliability and hampers robust downstream analyses. Here, we present a biologically grounded QC framework for chromatin accessibility bigWig files that integrates peak-level information, background noise estimation, and recovery of stable genomic reference features. Using an ML-based peak caller (LanceOtron), we derive complementary quality metrics capturing signal structure and signal-to-noise properties. We further define constant promoter and CTCF regions as internal biological controls and show that their recovery provides a sensitive measure of data quality across diverse cellular contexts. We apply this framework to a collection of 502 human chromatin accessibility bigWig tracks spanning a wide range of tissues and cell types. The proposed metrics capture related but non-redundant aspects of signal quality and motivate the use of constant promoter and CTCF recovery as biologically meaningful targets. An XGBoost model trained on LanceOtron-derived features accurately predicts recovery of these stable genomic elements on held-out data (R2 = 0.97), yielding a continuous and interpretable quality score. Feature importance analysis using SHAP values highlights that model decisions are driven by biologically relevant signal properties rather than arbitrary heuristics. Quantile-based stratification of the quality score is further supported by clear qualitative differences in genome browser visualisations. Together, this work provides a principled and extensible frame-work for assessing the quality of chromatin accessibility bigWig tracks, enabling more reliable data integration and supporting downstream ML applications in regulatory genomics.

bioinformatics↗

bioRxiv2026

CREP: Cis-Regulatory Element Predictor Based on Fine-Tuned Enformer

A substantial fraction of disease-associated genetic variants reside in non-coding regions of the genome, where they act by perturbing cis-regulatory elements (CREs) such as enhancers, promoters, and insulators. While recent sequence-based deep learning models, such as Enformer, accurately predict continuous epigenomic signals from DNA sequence, they do not directly provide discrete and interpretable CRE annotations. Here, we present CREP (Cis-Regulatory Element Predictor), a fine-tuned version of Enformer trained to predict regulatory element identity from sequence using REgulamentary-derived annotations across multiple human cell-types. Through a controlled experimental framework, we show that incorporating diverse cell-types improves model performance. CREP leverages cell-type-specific training data to learn regulatory representations while producing a unified prediction of CRE identity from sequence. This is demonstrated by the Vanuatu SNP, a non-coding variant that creates a de novo erythroid regulatory element, which is correctly detected only when erythroid data are included during training. Error analysis further reveals that apparent misclassifications between enhancers and promoters reflect their shared regulatory architecture, supporting the view of CREs as a functional continuum rather than strictly discrete classes. Together, these results demonstrate that CREP enables interpretable prediction of regulatory element identity from sequence and provides a framework for the functional interpretation of non-coding genetic variation.

bioinformatics↗

bioRxiv2026

Expert-Guided Supervised Annotation of Erythroid Differentiation in Single-Cell RNA-seq

Accurate annotation of intermediate cell states remains a major challenge in single-cell RNA sequencing (scRNA-seq), particularly in continuous differentiation systems such as erythropoiesis. Existing reference-based methods often lack the resolution required to distinguish early and transitional erythroid progenitors and may generalise poorly across datasets and modalities. Here, we present a supervised framework for erythroid lineage annotation based on expert-curated training data that integrates bulk and single-cell transcriptomic information. Starting from a human bone marrow scRNA-seq atlas, we refined erythroid annotations by introducing previously unresolved progenitor stages, including burst-forming unit-erythroid (BFU-E), colony-forming unit-erythroid (CFU-E), and pro-erythroblast (ProE), guided by canonical marker genes and bulk RNA-seq references. We trained and benchmarked four classical machine learning models and identified LightGBM as the best-performing approach, achieving a validation macro F1-score of 0.821 and balanced accuracy of 0.826. On a held-out test set, the model showed strong performance across most erythroid stages, with errors largely confined to adjacent differentiation states. The classifier was further transferred to independent bulk RNA-seq samples and an external bone marrow scRNA-seq dataset, where it recovered expected erythroid progression and refined coarse-grained annotations into higher-resolution cell states. Together, these results show that expert-curated supervised learning can improve erythroid cell state annotation in scRNA-seq and provide a practical framework for studying differentiation hierarchies in settings where finely resolved public references are limited.

bioinformatics↗

bioRxiv2026

Enformer-Based Phylogenetic Tree Reconstruction

Enformer is a deep learning model trained on human and mouse genomes to predict regulatory activity from 196,608 bp DNA windows. Its trunk embeddings capture long-range cis-regulatory interactions, but whether this signal generalises across the tree of life has not been assessed. We embed universal single-copy orthologous groups (OGs) from OrthoDB v12 across three taxonomic scales and evaluate reconstructed trees against TimeTree5 using Mantel r and Normalised Robinson-Foulds (NRF). On 702 OGs across 34 Primate species ([≤] 74 Mya), the consensus tree achieves Mantel r = 0.902 and NRF = 0.481, correctly recovering major clades. A key finding is that flanking regulatory context -- not the gene locus itself -- carries the phylogenetic signal: restricting pooling to central 448 bins collapses Mantel r to 0.355. Applying the same fixed configuration to Vertebrates ([≤] 450 Mya, 83 OGs, 150 species) and Plants ([≤] 1,500 Mya, 92 OGs, 40 species) yields consensus Mantel r of 0.752 and 0.803 respectively, with NRF worsening monotonically across tiers. Distance-ordering fidelity degrades smoothly with evolutionary distance while topological accuracy declines steadily, with no sharp taxonomic boundary. These results show that an unmodified regulatory deep learning model encodes robust phylogenetic signal well beyond its training distribution, reaching across 1,500 million years of divergence.

bioinformatics↗

bioRxiv2025

TRANCERs: Engineering enhancers into autonomous tissue-specific expression cassettes

Due to the scarcity of autonomous cell-specific promoters, control of transgene expression remains a major challenge, particularly for therapeutic applications where specificity and compactness are critical. We developed TRANCERs (TRanscriptionally Autonomous enhaNCERs), a modular system that converts gene regulatory elements (enhancers) into cell-specific standalone promoters by coupling enhancer activity with known and novel principles of transcript stability and nuclear export. This vastly increases the current repertoire of highly specific promoters available to science. We show TRANCERs drive lineage-restricted expression of genes both in vitro and in vivo, and that this expression pattern faithfully mirrors the activity of the enhancer element used. Their compact footprint and resistance to epigenetic silencing makes them ideally suited for payload-limited delivery systems such as viral vectors. TRANCERs provide a broadly applicable strategy for precise and programmable control of gene expression, enabling highly specific and customisable expression patterns for both basic research and gene-based therapies.

synthetic biology↗

bioRxiv2025

REnformer, a single-cell ATAC-seq predicting model to investigate open chromatin sites

Genome regulatory elements are fundamental to cellular identity and cell type specific gene expression. Understanding how the underlying genetic code is differentially utilised by different cell types is central to understanding human health and disease. To better understand how DNA encodes genome regulatory elements such as promoters, enhancers, and boundary elements, we leverage the Enformer gene expression and epigenetic prediction model. We used transfer learning with high quality single cell ATAC datasets to develop REnformer, a model to predict chromatin accessibility. By introducing a bench-mark for comparing performances against Enformer model, REnformer significantly outperformed Enformer in terms of higher prediction outcomes and lower error rates in all extensive analyses shown; introducing these benchmarks allowed us, and possible future works, to fairly compare such models. We further tested REnformer by predicting the effects of a well characterised -thalassemia variant and found that the prediction aligned with the observed change in genome regulatory element, previously validated. We conclude that REnformer is and can be a state-of-the-art tool to predict cell type specific regulatory elements, and interrogate the effect of genome variation in health and disease.

bioinformatics↗

bioRxiv2025

Sustained MYB activity drives emergent enhancer activation and precise enhancer-promoter interactions

Transcription factors (TFs) are key effectors of enhancer activity. MYB is a critical hematopoietic TF that is frequently dysregulated in cancer. Despite its well-established role, the exact mechanisms by which MYB influences enhancer function--and the specific stages of enhancer activation at which it operates--remain poorly understood. Using high resolution Micro-Capture-C, we show that upon MYB degradation, highly defined enhancer-promoter interactions at specific MYB binding sites are lost. Loss of these interactions, together with other hallmarks of enhancer activity--reduced H3 lysine-27 acetylation and enhancer RNA transcription--correlates with significant downregulation of target gene expression in leukemia, indicating that MYB mediates transcription activation via maintenance of enhancer function. When anchored to DNA within a gene desert region that is devoid of histone marks and active transcription, the MYB transactivation domain is sufficient and necessary for the nucleation of an enhancer-like region. This results in the activation of transcription from distal cryptic elements and the establishment of long-range chromatin interactions up to 400 kb away from the anchor point. Together, these results indicate that MYB activity alone is sufficient to induce long-range interactions and transcription, achieving this through highly precise enhancer-promoter crosstalk.

genomics↗

bioRxiv2024

Deciphering cis-regulatory elements using REgulamentary

With the boom in Genome-Wide Association Studies (GWAS), it has become apparent that many disease-associated genetic variants lie in the non-coding regions of the genome. In order to prioritise these variants and disentangle their functional significance, it is important to be able to accurately classify cis-regulatory elements within these non-coding regions of the genome. Historically, the classification of cis-regulatory elements relied purely on the presence of characteristic histone marks, with recent advancements in their classification using more sophisticated Hidden Markov Model (HMM)-based approaches. The limitation of the HMM-based approaches is that the output of these models is an arbitrary chromatin state, which then requires the user to manually assign these states to a particular class of cis-regulatory elements. Here we present a new tool, REgulamentary, which enables de novo genome-wide annotation of cis-regulatory elements in a cell-type specific manner. We benchmarked REgulamentary against GenoSTAN, the most popular existing published chromatin annotation and regulatory element identification tool, to demonstrate the advancements REgulamentary can provide in assigning chromatin states. Finally, as an example of REgulamentarys utility in solving complex disease trait loci, we applied REgulamentary to published GWAS data to demonstrate how this tool can be used to prioritise likely causal variants.

bioinformatics↗

bioRxiv2024

Focal Deletions of a Promoter Tether Activate the IRX3 Oncogene in T Cell Acute Lymphoblastic Leukemia

Oncogenes can be activated in cis through multiple mechanisms including enhancer hijacking events and noncoding mutations that create enhancers or promoters de novo. These paradigms have helped parse somatic variation of noncoding cancer genomes, thereby providing a rationale to identify noncanonical mechanisms of gene activation. Here we describe a novel mechanism of oncogene activation whereby focal copy number loss of an intronic element within the FTO gene leads to aberrant expression of IRX3, an oncogene in T cell acute lymphoblastic leukemia (T-ALL). Loss of this CTCF bound element downstream to IRX3 (+224 kb) leads to enhancer hijack of an upstream developmentally active super-enhancer of the CRNDE long noncoding RNA (-644 kb). Unexpectedly, the CRNDE super-enhancer interacts with the IRX3 promoter with no transcriptional output until it is untethered from the FTO intronic site. We propose that promoter tethering of oncogenes to inert regions of the genome is a previously unappreciated biological mechanism preventing tumorigenesis.

cancer biology↗

bioRxiv2023

Active regulatory elements recruit cohesin to establish cell-specific chromatin domains.

As the structure of the genome is analysed at ever increasing resolution it is becoming clear that there is considerable variation in the 3D chromatin architecture across different cell types. It has been proposed that this may, in part, be due to increased recruitment of cohesin to activated cis-elements (enhancers and promoters) leading to cell-type specific loop extrusion underlying the formation of new subTADs. Here we show that cohesin correlates well with the presence of active enhancers and this varies in an allele-specific manner with the presence or absence of polymorphic enhancers which vary from one individual to another. Using the alpha globin cluster as a model, we show that when all enhancers are removed, peaks of cohesin disappear from these regions and the erythroid specific subTAD is no longer formed. Re-insertion of the major alpha globin enhancer (R2) is associated with the appearance of a new peak of cohesin at the site of insertion. In complementary experiments insertion of R2 into a "neutral" region of the genome recruits cohesin, induces transcription and creates a new large (75kb) erythroid specific domain. Together these findings support the proposal that active enhancers recruit cohesin, stimulate loop extrusion and promote the formation of cell specific subTADs.

genomics↗

bioRxiv2023

Loop extrusion by cohesin plays a key role in enhancer-activated gene expression during differentiation

Enhancers and their target promoters often come into close physical proximity when activated. This proximity may be explained by a variety of mechanisms; most recently via cohesin-mediated chromatin loop extrusion. Despite this compelling hypothesis, acute depletion of cohesin does not cause widespread changes in gene expression. We have tested the role of cohesin-mediated loop extrusion on gene expression at the mouse alpha-globin locus during erythropoiesis. Acute depletion of cohesin downregulates alpha-globin expression at early but not late stages of differentiation. When single or multiple CTCF sites are placed between the alpha-globin enhancers and promoters, alpha-gene expression is downregulated. Importantly, the orientation of the CTCF site plays a critical role, suggesting that within this activated domain, cohesin predominantly but not exclusively translocates from the enhancers to the promoters. We find that loop extrusion does play an important role in establishing enhancer-promoter proximity and consequent expression of inducible genes during differentiation.

molecular biology↗

bioRxiv2023

Cross-species regulatory landscapes and elements revealed by novel joint systematic integration of human and mouse blood cell epigenomes

Knowledge of locations and activities of cis-regulatory elements (CREs) is needed to decipher basic mechanisms of gene regulation and to understand the impact of genetic variants on complex traits. Previous studies identified candidate CREs (cCREs) using epigenetic features in one species, making comparisons difficult between species. In contrast, we conducted an interspecies study defining epigenetic states and identifying cCREs in blood cell types to generate regulatory maps that are comparable between species, using integrative modeling of eight epigenetic features jointly in human and mouse in our Validated Systematic Integration (VISION) Project. The resulting catalogs of cCREs are useful resources for further studies of gene regulation in blood cells, indicated by high overlap with known functional elements and strong enrichment for human genetic variants associated with blood cell phenotypes. The contribution of each epigenetic state in cCREs to gene regulation, inferred from a multivariate regression, was used to estimate epigenetic state Regulatory Potential (esRP) scores for each cCRE in each cell type, which were used to categorize dynamic changes in cCREs. Groups of cCREs displaying similar patterns of regulatory activity in human and mouse cell types, obtained by joint clustering on esRP scores, harbored distinctive transcription factor binding motifs that were similar between species. An interspecies comparison of cCREs revealed both conserved and species-specific patterns of epigenetic evolution. Finally, we showed that comparisons of the epigenetic landscape between species can reveal elements with similar roles in regulation, even in the absence of genomic sequence alignment.

genomics↗

bioRxiv2022

Chromatin remodeller Chd7 is developmentally regulated in the neural crest by tissue-specific transcription factors

Neurocristopathies such as CHARGE syndrome result from aberrant neural crest development. A large proportion of CHARGE cases are attributed to mutations in the gene encoding CHD7, chromodomain helicase DNA binding protein 7, which remodels chromatin. While the role for CHD7 in neural crest development is well documented, it remains elusive how this seemingly ubiquitous factor is specifically upregulated in neural crest cells. Here, we use epigenomic profiling of chick neural crest to identify a cohort of enhancers regulating Chd7 expression in neural crest cells and other tissues. We functionally validate upstream transcription factor binding at candidate enhancers, revealing novel epistatic relationships between neural crest master regulators and Chd7. To our knowledge, this is the first report of tissue-specific regulation of a chromatin remodeller. Furthermore, we find conserved enhancer features in human embryonic epigenomic data and validate the activity of the human equivalent CHD7 enhancers in the chick embryo. Collectively our findings embed Chd7 in the neural crest gene regulatory network and offer potentially clinically relevant elements for interpreting CHARGE syndrome cases without causative allocation.

developmental biology↗

bioRxiv2022

Inferring causal genes at type 2 diabetes GWAS loci through chromosome interactions in islet cells

Resolving causal genes for type 2 diabetes at loci implicated by genome-wide association studies (GWAS) requires integrating functional genomic data from relevant cell types. Chromatin features in endocrine cells of the pancreatic islet are particularly informative and recent studies leveraging chromosome conformation capture (3C) with Hi-C based methods have elucidated regulatory mechanisms in human islets. However, these genome-wide approaches are less sensitive and afford lower resolution than methods that target specific loci. To gauge the extent to which targeted 3C further resolves chromatin-mediated regulatory mechanisms at GWAS loci, we generated interaction profiles at 23 loci using next-generation (NG) Capture-C in a human beta cell model (EndoC-{beta}H1) and contrasted these maps with Hi-C maps in EndoC-{beta}H1 cells and human islets and a promoter capture Hi-C map in human islets. We found improvements in assay sensitivity of up to 33-fold and resolved 4.8X more chromatin interactions. At a subset of 18 loci with 25 co-localised GWAS and eQTL signals, NG Capture-C interactions implicated effector transcripts at five additional genetic signals relative to promoter capture Hi-C through physical contact with gene promoters. Therefore, high resolution chromatin interaction profiles at selectively targeted loci can complement genome- and promoter-wide maps.

genetics↗

bioRxiv2022

Multipartite super-enhancers function in an orientation-dependent manner

Transcriptional enhancers regulate gene expression in a developmental-stage and cell-specific manner. They were originally defined as individual regulatory elements that activate expression regardless of distance and orientation to their cognate genes. Genome-wide studies have shown that the mammalian enhancer landscape is much more complex, with different classes of individual enhancers and clusters of enhancer-like elements combining in additive, synergistic and redundant manners, possibly acting as single, integrated regulatory elements. These so-called super-enhancers are largely defined as clusters of enhancer-like elements which recruit particularly high levels of Mediator and often drive high levels of expression of key lineage-specific genes. Here, we analysed 78 erythroid-specific super-enhancers and showed that, as units, they preferentially interact in a directional manner, to drive expression of their cognate genes. Using the well characterised -globin super-enhancer, we show that inverting this entire structure severely downregulates -globin expression and activates flanking genes 5 of the super-enhancer. Our detailed genetic dissection of the -globin locus clearly attributes the clusters functional directionality to its sequence orientation, demonstrating that, unlike regular enhancers, super-enhancers act in an orientation-dependent manner. Together, these findings identify a novel emergent property of super-enhancers and revise current models by which enhancers are thought to contact and activate their cognate genes.

molecular biology↗

bioRxiv2021

Chromatin interaction maps identify Wnt responsive cis-regulatory elements coordinating Paupar-Pax6 expression in neuronal cells

Central nervous system-expressed long non-coding RNAs (lncRNAs) are often located in the genome close to protein coding genes involved in transcriptional control. Such lncRNA-protein coding gene pairs are frequently temporally and spatially co-expressed in the nervous system and are predicted to act together to regulate neuronal development and function. Although some of these lncRNAs also bind and modulate the activity of the encoded transcription factors, the regulatory mechanisms controlling co-expression of neighbouring lncRNA-protein coding genes remain unclear. Here, we used high resolution NG Capture-C to map the cis-regulatory interaction landscape of the key neuro-developmental Paupar-Pax6 lncRNA-mRNA locus. The results defined chromatin architecture changes associated with high Paupar-Pax6 expression in neurons and identified both promoter selective as well as shared cis-regulatory interactions with the Paupar and Pax6 promoters involved in regulating Paupar-Pax6 co-expression in neuronal cells. The TCF7L2 transcription factor, a major regulator of chromatin architecture and effector of the Wnt signalling pathway, binds to a subset of these candidate cis-regulatory elements to coordinate Paupar and Pax6 co-expression. We identify a functional TCF7L2 bound cis-regulatory element within the Paupar gene, suggesting that the Paupar DNA locus itself regulates Pax6 expression in addition to its previously described transcriptdependent modes of action. Our work provides important insights into the chromatin interactions, signalling pathways and transcription factors controlling co-expression of adjacent lncRNAs and protein coding genes in the brain.

genomics↗

bioRxiv2021

Dynamic Runx1 chromatin boundaries affect gene expression in hematopoietic development

The transcription factor RUNX1 is a critical regulator of developmental hematopoiesis and is frequently disrupted in leukemia. Runx1 is a large, complex gene that is expressed from two alternative promoters under the spatiotemporal control of multiple hematopoietic enhancers. To dissect the dynamic regulation of Runx1 in hematopoietic development, we analyzed its three-dimensional chromatin conformation in mouse embryonic stem cell (ESC) differentiation cultures. Runx1 resides in a 1.1 Mb topologically associating domain (TAD) demarcated by convergent CTCF motifs. As ESCs differentiate to mesoderm, chromatin accessibility, Runx1 enhancer-promoter (E-P) interactions, and CTCF-CTCF interactions increased in the TAD, along with initiation of Runx1 expression from the P2 promoter. Differentiation to hematopoietic progenitor cells was associated with the formation of tissue-specific sub-TADs over Runx1, a shift in E-P interactions, P1 promoter demethylation, and robust expression from both Runx1 promoters. Deletions of promoter-proximal CTCF sites at the sub-TAD boundaries had no obvious effects on E-P interactions but led to partial loss of domain structure, mildly affected gene expression, and delayed hematopoietic development. Together, our analyses of gene regulation at a large multi-promoter developmental gene revealed that dynamic sub-TAD chromatin boundaries play a role in establishing TAD structure and coordinated gene expression.

developmental biology↗

bioRxiv2021

LanceOtron: a deep learning peak caller for ATAC-seq, ChIP-seq, and DNase-seq

ATAC-seq, ChIP-seq, and DNase-seq have revolutionized molecular biology by allowing researchers to identify important DNA-encoded elements genome-wide. Regions where these elements are found appear as peaks in the analog signal of an assays coverage track, and despite the ease with which humans can visually categorize these regions, meaningful peak calls from whole genome datasets require complex analytical techniques. Current methods focus on statistical tests to classify peaks, reducing the information-dense peak shapes to simply maximum height, and discounting that background signals do not completely follow any known probability distribution for significance testing. Deep learning has been shown to be highly accurate for image recognition, on par or exceeding human ability, providing an opportunity to reimagine and improve peak calling. We present the peak calling framework LanceOtron, which combines multifaceted enrichment measurements with deep learning image recognition techniques for assessing peak shape. In benchmarking transcription factor binding, chromatin modification, and open chromatin datasets, LanceOtron outperforms the long-standing, gold-standard peak caller MACS2 through its improved selectivity and near perfect sensitivity. In addition to command line accessibility, a graphical web application was designed to give any researcher the ability to generate optimal peak calls and interactive visualizations in a single step.

bioinformatics↗