Epigenomic Data Research Articles

Integrative 3D genomics with multi-omics analysis and functional validation of genetic regulatory mechanisms of abdominal fat deposition in chickens

Chickens are the most abundant agricultural animals globally, with controlling abdominal fat deposition being a key objective in poultry breeding. While GWAS can identify genetic variants associated with abdominal fat deposition, the precise roles and mechanisms of these variants remain largely unclear. Here, we use male chickens from two lines divergently selected for abdominal fat deposition as experimental models. Through the integration of genomic, epigenomic, 3D genomic, and transcriptomic data, we build a comprehensive chromatin 3D regulatory network map to identify the genetic regulatory mechanisms that influence abdominal fat deposition in chickens. Notably, we find that the rs734209466 variant functions as an allele-specific enhancer, remotely enhancing the transcription of IGFBP2 and IGFBP5 by the binding transcription factor IRF4. This interaction influences the differentiation and proliferation of preadipocytes, which ultimately affects phenotype. This work presents a detailed genetic regulatory map for chicken abdominal fat deposition, offering molecular targets for selective breeding.

Predicting protein synergistic effect in Arabidopsis using epigenome profiling

Histone modifications can regulate transcription epigenetically by marking specific genomic loci, which can be mapped using chromatin immunoprecipitation sequencing (ChIP-seq). Here we present QHistone, a predictive database of 1534 ChIP-seqs from 27 histone modifications in Arabidopsis, offering three key functionalities. Firstly, QHistone employs machine learning to predict the epigenomic profile of a query protein, characterized by its most associated histone modifications, and uses these modifications to infer the protein’s role in transcriptional regulation. Secondly, it predicts synergistic regulatory activities between two proteins by comparing their profiles. Lastly, it detects previously unexplored co-regulating protein pairs by screening all known proteins. QHistone accurately identifies histone modifications associated with specific known proteins, and allows users to computationally validate their results using gene expression data from various plant tissues. These functions demonstrate an useful approach to utilizing epigenome data for gene regulation analysis, making QHistone a valuable resource for the scientific community (https://qhistone.paoyang.ipmb.sinica.edu.tw).

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 pathogenic variants 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, how this factor is specifically up-regulated in neural crest cells is not understood. Here, we use epigenomic profiling of chick and human 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, showing tissue-specific regulation of a globally acting 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. 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.

Best practices for differential accessibility analysis in single-cell epigenomics

Differential accessibility (DA) analysis of single-cell epigenomics data enables the discovery of regulatory programs that establish cell type identity and steer responses to physiological and pathophysiological perturbations. While many statistical methods to identify DA regions have been developed, the principles that determine the performance of these methods remain unclear. As a result, there is no consensus on the most appropriate statistical methods for DA analysis of single-cell epigenomics data. Here, we present a systematic evaluation of statistical methods that have been applied to identify DA regions in single-cell ATAC-seq (scATAC-seq) data. We leverage a compendium of scATAC-seq experiments with matching bulk ATAC-seq or scRNA-seq in order to assess the accuracy, bias, robustness, and scalability of each statistical method. The structure of our experiments also provides the opportunity to define best practices for the analysis of scATAC-seq data beyond DA itself. We leverage this understanding to develop an R package implementing these best practices.

BIT: Bayesian Identification of Transcriptional Regulators from Epigenomics-Based Query Region Sets.

Transcriptional regulators (TRs) are master controllers of gene expression and play a critical role in both normal tissue development and disease progression. However, existing computational methods for identification of TRs regulating specific biological processes have significant limitations, such as relying on distance on a linear chromosome or binding motifs that have low specificity. Many also use statistical tests in ways that lack interpretability and rigorous confidence measures. We introduce BIT, a novel Bayesian hierarchical model for in-silico TR identification. Leveraging a comprehensive library of TR ChIP-seq data, BIT offers a fully integrated Bayesian approach to assess genome-wide consistency between user-provided epigenomic profiling data and the TR binding library, enabling the identification of critical TRs while quantifying uncertainty. It avoids estimation and inference in a sequential manner or numerous isolated statistical tests, thereby enhancing accuracy and interpretability. BIT successfully identified critical TRs in perturbation experiments, functionally essential TRs in various cancer types, and cell-type-specific TRs within heterogeneous cell populations, offering deeper biological insights into transcriptional regulation.

SNPs in cytochromes P450 catalyzing cholesterol degradation in brain are associated with Parkinson's disease.

Besides being an essential structural component of plasma membranes and the precursor of many functional compounds and signaling molecules, cholesterol was also proposed to play a role in the etiology and/or manifestation of Parkinson's disease (PD). However, so far systematic investigations on the role of cholesterol and its metabolites present in the brain for the etiology of PD are missing. Here, we investigate for the first time the association of PD with SNPs in the genes of four cytochromes P450 (P450), CYP46A1, CYP39A1, CYP27A1 and CYP7B1, which are critical for the degradation of cholesterol in the brain. Analyzing 1,349 individuals from the PPMI data base, we found 24 SNPs in these four genes, which are significantly over- or under-represented in patients suffering from idiopathic PD (IPD). Studying each of the 362 IPD patients individually, we found that most patients (45%) showed only one associated SNP in one of the four P450 genes, while 31% displayed two associated SNPs and 18% three associated SNPs. The occurrence of some associated SNPs is in the same order of magnitude as SNPs in the GBA (beta-glucocerebrosidase) and thus might reflect a genetic predisposition for PD. As all 24 SNPs were located in introns and 3' untranslated regions, we evaluated the prospective regulatory impact of the surrounding genomic regions by using transcriptome and epigenome data from the Foundational Data Initiative for Parkinson Disease (FOUNDIN-PD). FOUNDIN-PD provides gene expression, open chromatin and DNA methylation data in a cohort of 89 induced pluripotent stem cell (iPSC) lines differentiated to dopaminergic (DA) neurons derived from people in the PPMI study. Indeed, two of the 24 SNPs, one in CYP7B1 (rs118111353) and the other one in CYP27A1 (rs74446825), were localized within a region of open chromatin in differentiated neurons. Interestingly, all iPSC lines with open chromatin in rs118111353 showed the reference allele. As all four P450, CYP46A1, CYP39A1, CYP27A1 and CYP7B1, are expressed in dopaminergic neurons, we discuss further functional studies to connect SNPs in regulatory regions with gene expression levels. Finally, potential possibilities for personalized therapeutic treatment of patients with SNPs in the four investigated P450 are discussed.

Contemporary Update on Clinical and Experimental Prostate Cancer Biomarkers: A Multi-Omics-Focused Approach to Detection and Risk Stratification.

Prostate cancer remains a significant health challenge, being the most prevalent non-cutaneous cancer in men worldwide. This review discusses the critical advancements in biomarker discovery using single-omics and multi-omics approaches. Multi-omics, integrating genomic, transcriptomic, proteomic, metabolomic, and epigenomic data, offers a comprehensive understanding of the molecular heterogeneity of prostate cancer, leading to the identification of novel biomarkers and therapeutic targets. This holistic approach not only enhances the specificity and sensitivity of prostate cancer detection but also supports the development of personalized treatment strategies. Key studies highlighted include the identification of novel genes, genetic mutations, peptides, metabolites, and potential biomarkers through multi-omics analyses, which have shown promise in improving prostate cancer management. The integration of multi-omics in clinical practice can potentially revolutionize prostate cancer prognosis and treatment, paving the way for precision medicine. This review underscores the importance of continued research and the application of multi-omics to overcome current challenges in prostate cancer diagnosis and therapy.

Epigenomic, transcriptomic and proteomic characterizations of reference samples.

A variety of newly developed next-generation sequencing technologies are making their way rapidly into the research and clinical applications, for which accuracy and cross-lab reproducibility are critical, and reference standards are much needed. Our previous multicenter studies under the SEQC-2 umbrella using a breast cancer cell line with paired B-cell line have produced a large amount of different genomic data including whole genome sequencing (Illumina, PacBio, Nanopore), HiC, and scRNA-seq with detailed analyses on somatic mutations, single-nucleotide variations (SNVs), and structural variations (SVs). However, there is still a lack of well-characterized reference materials which include epigenomic and proteomic data. Here we further performed ATAC-seq, Methyl-seq, RNA-seq, and proteomic analyses and provided a comprehensive catalog of the epigenomic landscape, which overlapped with the transcriptomes and proteomes for the two cell lines. We identified >7,700 peptide isoforms, where the majority (95%) of the genes had a single peptide isoform. Protein expression of the transcripts overlapping CGIs were much higher than the protein expression of the non-CGI transcripts in both cell lines. We further demonstrated the evidence that certain SNVs were incorporated into mutated peptides. We observed that open chromatin regions had low methylation which were largely regulated by CG density, where CG-rich regions had more accessible chromatin, low methylation, and higher gene and protein expression. The CG-poor regions had higher repressive epigenetic regulations (higher DNA methylation) and less open chromatin, resulting in a cell line specific methylation and gene expression patterns. Our studies provide well-defined reference materials consisting of two cell lines with genomic, epigenomic, transcriptomic, scRNA-seq and proteomic characterizations which can serve as standards for validating and benchmarking not only on various omics assays, but also on bioinformatics methods. It will be a valuable resource for both research and clinical communities.

Improved inference of population histories by integrating genomic and epigenomic data

Improved inference of population histories by integrating genomic and epigenomic data

Improved inference of population histories by integrating genomic and epigenomic data.

With the availability of high-quality full genome polymorphism (SNPs) data, it becomes feasible to study the past demographic and selective history of populations in exquisite detail. However, such inferences still suffer from a lack of statistical resolution for recent, for example bottlenecks, events, and/or for populations with small nucleotide diversity. Additional heritable (epi)genetic markers, such as indels, transposable elements, microsatellites, or cytosine methylation, may provide further, yet untapped, information on the recent past population history. We extend the Sequential Markovian Coalescent (SMC) framework to jointly use SNPs and other hyper-mutable markers. We are able to (1) improve the accuracy of demographic inference in recent times, (2) uncover past demographic events hidden to SNP-based inference methods, and (3) infer the hyper-mutable marker mutation rates under a finite site model. As a proof of principle, we focus on demographic inference in Arabidopsis thaliana using DNA methylation diversity data from 10 European natural accessions. We demonstrate that segregating single methylated polymorphisms (SMPs) satisfy the modeling assumptions of the SMC framework, while differentially methylated regions (DMRs) are not suitable as their length exceeds that of the genomic distance between two recombination events. Combining SNPs and SMPs while accounting for site- and region-level epimutation processes, we provide new estimates of the glacial age bottleneck and post-glacial population expansion of the European A. thaliana population. Our SMC framework readily accounts for a wide range of heritable genomic markers, thus paving the way for next-generation inference of evolutionary history by combining information from several genetic and epigenetic markers.

Review and Evaluate the Bioinformatics Analysis Strategies of ATAC-seq and CUT&Tag Data.

Efficient and reliable profiling methods are essential to study epigenetics. Tn5, one of the first identified prokaryotic transposases with high DNA-binding and tagmentation efficiency, is widely adopted in different genomic and epigenomic protocols for high-throughputly exploring the genome and epigenome. Based on Tn5, the Assay for Transposase-Accessible Chromatin using sequencing (ATAC-seq) and the Cleavage Under Targets and Tagmentation (CUT&Tag) were developed to measure chromatin accessibility and detect DNA-protein interactions. These methodologies can be applied to large amounts of biological samples with low-input levels, such as rare tissues, embryos, and sorted single cells. However, fast and proper processing of these epigenomic data has become a bottleneck because massive data production continues to increase quickly. Furthermore, inappropriate data analysis can generate biased or misleading conclusions. Therefore, it is essential to evaluate the performance of Tn5-based ATAC-seq and CUT&Tag data processing bioinformatics tools, many of which were developed mostly for analyzing chromatin immunoprecipitation followed by sequencing (ChIP-seq) data. Here, we conducted a comprehensive benchmarking analysis to evaluate the performance of eight popular software for processing ATAC-seq and CUT&Tag data. We compared the sensitivity, specificity, and peak width distribution for both narrow-type and broad-type peak calling. We also tested the influence of the availability of control IgG input in CUT&Tag data analysis. Finally, we evaluated the differential analysis strategies commonly used for analyzing the CUT&Tag data. Our study provided comprehensive guidance for selecting bioinformatics tools and recommended analysis strategies, which were implemented into Docker/Singularity images for streamlined data analysis.

Advances in liquid biopsy: Computational and artificial intelligence approaches in cancer research

Liquid biopsy analysis has emerged as a promising approach for non-invasive cancer monitoring and diagnosis. This review provides an overview of the current landscape and future potential of liquid biopsy in cancer research, with a particular focus on the computational methodologies and techniques utilized for liquid biopsy data analysis and interpretation. The challenges and opportunities in extracting meaningful insights from the vast array of genomic, epigenomic, transcriptomic, and proteomic data are discussed, as well as the possibilities and current pitfalls of artificial intelligence approaches. In addition, the benefits and limitations of integrating multimodal cancer research data are covered, with a focus on advancing precision oncology and personalized medicine. By providing a critical assessment of the field, this review aims to foster knowledge about the available computational approaches and facilitate the choice of the most appropriate methodology for in silico investigation of liquid biopsy data, ultimately enhancing research endeavors and disease management strategies. Most of the works discussed in this review have emerged within the past 5 years, indicating a rapidly growing interest in this technique.

The Genetic Determinants and Genomic Consequences of Non-Leukemogenic Somatic Point Mutations.

Clonal hematopoiesis (CH) is defined by the expansion of a lineage of genetically identical cells in blood. Genetic lesions that confer a fitness advantage, such as point mutations or mosaic chromosomal alterations (mCAs) in genes associated with hematologic malignancy, are frequent mediators of CH. However, recent analyses of both single cell-derived colonies of hematopoietic cells and population sequencing cohorts have revealed CH frequently occurs in the absence of known driver genetic lesions. To characterize CH without known driver genetic lesions, we used 51,399 deeply sequenced whole genomes from the NHLBI TOPMed sequencing initiative to perform simultaneous germline and somatic mutation analyses among individuals without leukemogenic point mutations (LPM), which we term CH-LPMneg. We quantified CH by estimating the total mutation burden. Because estimating somatic mutation burden without a paired-tissue sample is challenging, we developed a novel statistical method, the Genomic and Epigenomic informed Mutation (GEM) rate, that uses external genomic and epigenomic data sources to distinguish artifactual signals from true somatic mutations. We performed a genome-wide association study of GEM to discover the germline determinants of CH-LPMneg. After fine-mapping and variant-to-gene analyses, we identified seven genes associated with CH-LPMneg (TCL1A, TERT, SMC4, NRIP1, PRDM16, MSRA, SCARB1), and one locus associated with a sex-associated mutation pathway (SRGAP2C). We performed a secondary analysis excluding individuals with mCAs, finding that the genetic architecture was largely unaffected by their inclusion. Functional analyses of SMC4 and NRIP1 implicated altered HSC self-renewal and proliferation as the primary mediator of mutation burden in blood. We then performed comprehensive multi-tissue transcriptomic analyses, finding that the expression levels of 404 genes are associated with GEM. Finally, we performed phenotypic association meta-analyses across four cohorts, finding that GEM is associated with increased white blood cell count and increased risk for incident peripheral artery disease, but is not significantly associated with incident stroke or coronary disease events. Overall, we develop GEM for quantifying mutation burden from WGS without a paired-tissue sample and use GEM to discover the genetic, genomic, and phenotypic correlates of CH-LPMneg.

Accurate identification of genes associated with brain disorders by integrating heterogeneous genomic data into a Bayesian framework

Genome-wide association studies (GWAS) have revealed many brain disorder-associated SNPs residing in the noncoding genome, rendering it a challenge to decipher the underlying pathogenic mechanisms. Here, we present an unsupervised Bayesian framework to identify disease-associated genes by integrating risk SNPs with long-range chromatin interactions (iGOAT), including SNP-SNP interactions extracted from ∼500,000 patients and controls from the UK Biobank, and enhancer-promoter interactions derived from multiple brain cell types at different developmental stages. The application of iGOAT to three psychiatric disorders and three neurodegenerative/neurological diseases predicted sets of high-risk (HRGs) and low-risk (LRGs) genes for each disorder. The HRGs were enriched in drug targets, and exhibited higher expression during prenatal brain developmental stages than postnatal stages, indicating their potential to affect brain development at an early stage. The HRGs associated with Alzheimer's disease were found to share genetic architecture with schizophrenia, bipolar disorder and major depressive disorder according to gene co-expression module analysis and rare variants analysis. Comparisons of this method to the eQTL-based method, the TWAS-based method, and the gene-level GWAS method indicated that the genes identified by our method are more enriched in known brain disorder-related genes, and exhibited higher precision. Finally, the method predicted 205 risk genes not previously reported to be associated with any brain disorder, of which one top-risk gene, MLH1, was experimentally validated as being schizophrenia-associated. iGOAT can successfully leverage epigenomic data, phenotype-genotype associations, and protein-protein interactions to advance our understanding of brain disorders, thereby facilitating the development of new therapeutic approaches. The work was funded by the National Key Research and Development Program of China (2024YFF1204902), the Natural Science Foundation of China (82371482), Guangzhou Science and Technology Research Plan (2023A03J0659) and Natural Science Foundation of Guangdong (2024A1515011363).

60 Dissecting the Cellular and Epigenetic Landscape of Clear Cell Renal Cell Carcinoma Through Paired snRNA/ATAC-Seq Analysis

Abstract Background Clear cell renal cell carcinoma (ccRCC) stands as one of the most prevalent and aggressive forms of renal malignancies, accounting for a substantial portion of kidney cancer-related mortality worldwide. The heterogeneity within ccRCC poses a significant challenge to understanding its molecular mechanisms and devising effective therapeutic strategies. [1] Previous studies utilizing single nucleus RNA sequencing (snRNA-seq) have revealed transcriptional contributions to cell-type specificity in both mature human and mouse kidneys. Moreover, recent advancements have expanded this approach to single-cell profiling of chromatin accessibility. Single-nuclei assay for transposase-accessible chromatin using sequencing (snATAC-seq) has enabled the measurement of chromatin accessibility in thousands of individual cells, providing insights into the dynamic process that drives nephron development and differentiation. Integration and analysis of multimodal single-cell datasets, such as snRNA-seq and snATAC-seq, allows for the prediction of cell-type-specific cis-regulatory DNA interactions and transcription factor activity, complementing the transcriptional information obtained by snRNA-seq.[2] [3] [1] Yang, Wang, and Yang, “Treatment Strategies for Clear Cell Renal Cell Carcinoma.” [2] Muto et al., “Single Cell Transcriptional and Chromatin Accessibility Profiling Redefine Cellular Heterogeneity in the Adult Human Kidney.” [3] Monteagudo-Sánchez, Noordermeer, and Greenberg, “The Impact of DNA Methylation on CTCF-Mediated 3D Genome Organization.” Methods This study leveraged paired snRNA/ATAC-seq in tumor and normal-adjacent samples from the Renal Tumor Biobank at Dartmouth (n= 63) to explore the ccRCC landscape at single-cell resolution using 10x multiome technology. The snRNA-seq dataset was processed with Seurat v5.0 to remove low-quality nuclei. Doublets were removed with DoubletFinder v2.02, and ambient DNA was removed using decontX from celda. Normalization was conducted via SCTransform, and missing data was imputed with ALRA. Clustering was performed by constructing a KNN graph and running the Louvain algorithm. Cell type assignments were predicted using Azimuth with the human kidney reference dataset. The snATAC-seq data was processed with ArchR v1.0.1, which performed dimensionality reduction and clustering. Cluster identities defined in the Seurat workflow were cross-annotated to the ATAT-seq data, and open peaks were identified within these cell-type clusters using MACS3 v3.0.1. Identification of marker peaks, motif enrichment, and motif enrichment deviations were inferred utilizing ArchR’s getmarkerFeatures, peakAnnoEnrichment, and addMotifAnnotations functions, respectively. Results Motif enrichment analysis revealed canonical cancer-associated transcription factor (TF) dysregulation in tumor cells (nephron loop-like), including enrichment of HNF1B, SMARCC1, FOSL1, and JUND. Interestingly, HNF and SMARCC were also enriched in the lymphoid compartment, indicating possible crosstalk between these two cell types. Previous work has investigated the involvement of the HNF1 in ccRCC and in chromophobe RCC for prognosis,and we plan on investigating the relationship of these enrichments to tumor grade, stage and outcome.[1] Additional TF enrichments included BCL families and ELF within the myeloid compartment, and vascularization related TF families in endothelial cell types including enrichment of various ETS TFs. Investigation of cell-cell interactions between immune and endothelial cell populations as well as between tumor and normal adjacent tissues will be performed to further disentangle the role differentially accessable chromatin plays in ccRCC. [1] Buchner et al., “Downregulation of HNF-1B in Renal Cell Carcinoma Is Associated With Tumor Progression and Poor Prognosis.” Conclusions Our multi-omic approach has several advantages. First, it enables the characterization of cellular heterogeneity within the tumor microenvironment, identifying cell populations and their contributions to disease progression. Secondly, by integrating transcriptomic and epigenomic data, we gain insights into the dynamic interplay between gene expression and chromatin accessibility, uncovering potential therapeutic vulnerabilities and biomarkers for personalized treatment strategies. DOD CDMRP Funding: yes

71 Exploring DNA-methylation and DNA-hydroxymethylation in clear cell renal cell carcinoma using Dirichlet multinomial regression

Abstract Background DNA methylation alterations have been found related to clear cell renal cell carcinoma (ccRCC). In addition, DNA methylation has been used to identify cell heterogeneity at the tumor level using cell deconvolution methods. Other cytosine modifications are less studied in the field. In this study, we aim to explore the use of compositional method, Dirichlet multinomial regression1, to simultaneously explore DNA methylation and DNA hydroxymethylation in ccRCC. Methods A total of 243 clear cell renal cell carcinoma (ccRCC) samples were collected from the Dartmouth Renal Tumor Biobank. DNA methylation analysis was conducted using the Infinium MethylationEPIC BeadChip, using tandem bisulfite (BS) and oxidative bisulfite (oxBS) treatments to differentiate between 5-methylcytosine (5mC) and 5-hydroxymethylcytosine (5hmC) states. Beta values were calculated for all the arrayed samples. Quality control was performed using the ENmix pipeline to ensure the inclusion of high-quality samples using a stringent pOOBHA <0.05. The oxBS maximum likelihood estimate (oxBS-MLE) method was applied to assess 5mC and 5hmC levels based on paired BS and oxBS datasets. To determine the inclusion threshold for 5hmC beta values in the Dirichlet analysis, the R package oxBSCut was used to mask values close to zero2. Missing 5hmC values and those values masked below the detection limit were imputed using a compositional approach and a Tobit regression model to retain the model properties. The Dirichlet model was applied to analyze three methylation states: unmethylated, methylated, and hydroxymethylated, providing comprehensive insights into the methylation landscape of ccRCC samples. Results Here, we analyzed the top 1% most variable methylated sites (n=3020). When comparing tumor vs normal adjacent samples we tested the cytosine modifications in the same model adjusting for sex, age at diagnosis, tumor stage, and grade. For this exploratory analysis, we selected those results that passed an FDR<0.01 and an absolute difference >0.2 for further exploration. Four sites were hyperhydroxymethylated related to MYO5A, DLG2, ANKRD33B, and one open sea site. A total of 23 sites were significantly hypermethylated. Using eForge-TF, these sites tracked to transcription factors targeting estrogen receptors, retinoic acid receptors alpha and gamma, and FOXC1 among others3. Conclusions In this preliminary analysis, using a new modeling approach to interrogate cytosine modifications simultaneously, we found some intriguing results consistent with previous literature. We will expand this analysis to evaluate the relation between extreme changes in hydroxymethylation in particular in relation to alternative splicing alterations. Reference 1. Tsagris, M., Stewart, C. A Dirichlet Regression Model for Compositional Data with Zeros. Lobachevskii J Math39, 398–412 (2018). https://doi.org/10.1134/S1995080218030198 2. Zhang Z, Lee MK, Perreard L, Kelsey KT, Christensen BC, Salas LA. Navigating the hydroxymethylome: experimental biases and quality control tools for the tandem bisulfite and oxidative bisulfite Illumina microarrays. Epigenomics. 2022;14(3):139-152. doi:10.2217/epi-2021-0490 3. Charles E Breeze, Alex P Reynolds, Jenny van Dongen, Ian Dunham, John Lazar, Shane Neph, Jeff Vierstra, Guillaume Bourque, Andrew E Teschendorff, John A Stamatoyannopoulos, Stephan Beck, eFORGE v2.0: updated analysis of cell type-specific signal in epigenomic data, Bioinformatics, Volume 35, Issue 22, November 2019, Pages 4767–4769, https://doi.org/10.1093/bioinformatics/btz456

EPInformer: a scalable deep learning framework for gene expression prediction by integrating promoter-enhancer sequences with multimodal epigenomic data.

Transcriptional regulation, critical for cellular differentiation and adaptation to environmental changes, involves coordinated interactions among DNA sequences, regulatory proteins, and chromatin architecture. Despite extensive data from consortia like ENCODE, understanding the dynamics of cis-regulatory elements (CREs) in gene expression remains challenging. Deep learning is a powerful tool for learning gene expression and epigenomic signals from DNA sequences, exhibiting superior performance compared to conventional machine learning approaches. However, even the most advanced deep learning-based methods may fall short in capturing the regulatory effects of distal elements such as enhancers, limiting their predictive accuracy. In addition, these methods may require significant resources to train or to adapt to newly generated data. To address these challenges, we present EPInformer, a scalable deep-learning framework for predicting gene expression by integrating promoter-enhancer interactions with their sequences, epigenomic signals, and chromatin contacts. Our model outperforms existing gene expression prediction models in rigorous cross-chromosome validation, accurately recapitulates enhancer-gene interactions validated by CRISPR perturbation experiments, and identifies crucial transcription factor motifs within regulatory sequences. EPInformer is available as open-source software at https://github.com/pinellolab/EPInformer.

Integration of multi-modal datasets to estimate human aging

Aging involves complex biological processes leading to the decline of living organisms. As population lifespan increases worldwide, the importance of identifying factors underlying healthy aging has become critical. Integration of multi-modal datasets is a powerful approach for the analysis of complex biological systems, with the potential to uncover novel aging biomarkers. In this study, we leveraged publicly available epigenomic, transcriptomic and telomere length data along with histological images from the Genotype-Tissue Expression project to build tissue-specific regression models for age prediction. Using data from two tissues, lung and ovary, we aimed to compare model performance across data modalities, as well as to assess the improvement resulting from integrating multiple data types. Our results demostrate that methylation outperformed the other data modalities, with a mean absolute error of 3.36 and 4.36 in the test sets for lung and ovary, respectively. These models achieved lower error rates when compared with established state-of-the-art tissue-agnostic methylation models, emphasizing the importance of a tissue-specific approach. Additionally, this work has shown how the application of Hierarchical Image Pyramid Transformers for feature extraction significantly enhances age modeling using histological images. Finally, we evaluated the benefits of integrating multiple data modalities into a single model. Combining methylation data with other data modalities only marginally improved performance likely due to the limited number of available samples. Combining gene expression with histological features yielded more accurate age predictions compared with the individual performance of these data types. Given these results, this study shows how machine learning applications can be extended to/in multi-modal aging research. Code used is available at https://github.com/zroger49/multi_modal_age_prediction.

A single-cell multimodal view on gene regulatory network inference from transcriptomics and chromatin accessibility data.

Eukaryotic gene regulation is a combinatorial, dynamic, and quantitative process that plays a vital role in development and disease and can be modeled at a systems level in gene regulatory networks (GRNs). The wealth of multi-omics data measured on the same samples and even on the same cells has lifted the field of GRN inference to the next stage. Combinations of (single-cell) transcriptomics and chromatin accessibility allow the prediction of fine-grained regulatory programs that go beyond mere correlation of transcription factor and target gene expression, with enhancer GRNs (eGRNs) modeling molecular interactions between transcription factors, regulatory elements, and target genes. In this review, we highlight the key components for successful (e)GRN inference from (sc)RNA-seq and (sc)ATAC-seq data exemplified by state-of-the-art methods as well as open challenges and future developments. Moreover, we address preprocessing strategies, metacell generation and computational omics pairing, transcription factor binding site detection, and linear and three-dimensional approaches to identify chromatin interactions as well as dynamic and causal eGRN inference. We believe that the integration of transcriptomics together with epigenomics data at a single-cell level is the new standard for mechanistic network inference, and that it can be further advanced with integrating additional omics layers and spatiotemporal data, as well as with shifting the focus towards more quantitative and causal modeling strategies.

Epigenetic control of metabolic identity across cell types.

Constraint-based network modelling is a powerful tool for analysing cellular metabolism at genomic scale. Here, we conducted an integrative analysis of metabolic networks reconstructed from RNA-seq data with paired epigenomic data from the EpiATLAS resource of the International Human Epigenome Consortium (IHEC). Applying a state-of-the-art contextualisation algorithm, we reconstructed metabolic networks across 1,555 samples corresponding to 58 tissues and cell types. Analysis of these networks revealed the distribution of metabolic functionalities across human cell types and provides a compendium of human metabolic activity. This integrative approach allowed us to define, across tissues and cell types, i) reactions that fulfil the basic metabolic processes (core metabolism), and ii) cell type-specific functions (unique metabolism), that shape the metabolic identity of a cell or a tissue. Integration with EpiATLAS-derived cell type-specific gene-level chromatin states and enhancer-gene interactions identified enhancers, transcription factors, and key nodes controlling core and unique metabolism. Transport and first reactions of pathways were enriched for high expression, active chromatin state, and Polycomb-mediated repression in cell types where pathways are inactive, suggesting that key nodes are targets of repression. This integrative analysis forms the basis for identifying regulation points that control metabolic identity in human cells.