Robust enhancer-gene regulation identified by single-cell transcriptomes and epigenomes.

Single-cell sequencing could help to solve the fundamental challenge of linking millions of cell-type-specific enhancers with their target genes. However, this task is confounded by patterns of gene co-expression in much the same way that genetic correlation due to linkage disequilibrium confounds fine-mapping in genome-wide association studies (GWAS). We developed a non-parametric permutation-based procedure to establish stringent statistical criteria to control the risk of false-positive associations in enhancer-gene association studies (EGAS). We applied our procedure to large-scale transcriptome and epigenome data from multiple tissues and species, including the mouse and human brain, to predict enhancer-gene associations genome wide. We tested the functional validity of our predictions by comparing them with chromatin conformation data and causal enhancer perturbation experiments. Our study shows how controlling for gene co-expression enables robust enhancer-gene linkage using single-cell sequencing data.

Enhlink infers distal and context-specific enhancer-promoter linkages.

Enhancers play a crucial role in regulating gene expression and their functional status can be queried with cell type precision using using single-cell (sc)ATAC-seq. To facilitate analysis of such data, we developed Enhlink, a novel computational approach that leverages single-cell signals to infer linkages between regulatory DNA sequences, such as enhancers and promoters. Enhlink uses an ensemble strategy that integrates cell-level technical covariates to control for batch effects and biological covariates to infer robust condition-specific links and their associated p-values. It can integrate simultaneous gene expression and chromatin accessibility measurements of individual cells profiled by multi-omic experiments for increased specificity. We evaluated Enhlink using simulated and real scATAC-seq data, including those paired with physical enhancer-promoter links enumerated by promoter capture Hi-C and with multi-omic scATAC-/RNA-seq data we generated from the mouse striatum. These examples demonstrated that our method outperforms popular alternative strategies. In conjunction with eQTL analysis, Enhlink revealed a putative super-enhancer regulating key cell type-specific markers of striatal neurons. Taken together, our analyses demonstrate that Enhlink is accurate, powerful, and provides features that can lead to novel biological insights.

Human-gained heart enhancers are associated with species-specific cardiac attributes.

The heart, a vital organ which is first to develop, has adapted its size, structure and function in order to accommodate the circulatory demands for a broad range of animals. Although heart development is controlled by a relatively conserved network of transcriptional/chromatin regulators, how the human heart has evolved species-specific features to maintain adequate cardiac output and function remains to be defined. Here, we show through comparative epigenomic analysis the identification of enhancers and promoters that have gained activity in humans during cardiogenesis. These cis-regulatory elements (CREs) are associated with genes involved in heart development and function, and may account for species-specific differences between human and mouse hearts. Supporting these findings, genetic variants that are associated with human cardiac phenotypic/disease traits, particularly those differing between human and mouse, are enriched in human-gained CREs. During early stages of human cardiogenesis, these CREs are also gained within genomic loci of transcriptional regulators, potentially expanding their role in human heart development. In particular, we discovered that gained enhancers in the locus of the early human developmental regulator ZIC3 are selectively accessible within a subpopulation of mesoderm cells which exhibits cardiogenic potential, thus possibly extending the function of ZIC3 beyond its conserved left-right asymmetry role. Genetic deletion of these enhancers identified a human gained enhancer that was required for not only ZIC3 and early cardiac gene expression at the mesoderm stage but also cardiomyocyte differentiation. Overall, our results illuminate how human gained CREs may contribute to human-specific cardiac attributes, and provide insight into how transcriptional regulators may gain cardiac developmental roles through the evolutionary acquisition of enhancers.

A single-cell atlas of chromatin accessibility in the human genome.

Current catalogs of regulatory sequences in the human genome are still incomplete and lack cell type resolution. To profile the activity of gene regulatory elements in diverse cell types and tissues in the human body, we applied single-cell chromatin accessibility assays to 30 adult human tissue types from multiple donors. We integrated these datasets with previous single-cell chromatin accessibility data from 15 fetal tissue types to reveal the status of open chromatin for ∼1.2 million candidate cis-regulatory elements (cCREs) in 222 distinct cell types comprised of >1.3 million nuclei. We used these chromatin accessibility maps to delineate cell-type-specificity of fetal and adult human cCREs and to systematically interpret the noncoding variants associated with complex human traits and diseases. This rich resource provides a foundation for the analysis of gene regulatory programs in human cell types across tissues, life stages, and organ systems.

DNA methylation atlas of the mouse brain at single-cell resolution.

Mammalian brain cells show remarkable diversity in gene expression, anatomy and function, yet the regulatory DNA landscape underlying this extensive heterogeneity is poorly understood. Here we carry out a comprehensive assessment of the epigenomes of mouse brain cell types by applying single-nucleus DNA methylation sequencing1,2 to profile 103,982 nuclei (including 95,815 neurons and 8,167 non-neuronal cells) from 45 regions of the mouse cortex, hippocampus, striatum, pallidum and olfactory areas. We identified 161 cell clusters with distinct spatial locations and projection targets. We constructed taxonomies of these epigenetic types, annotated with signature genes, regulatory elements and transcription factors. These features indicate the potential regulatory landscape supporting the assignment of putative cell types and reveal repetitive usage of regulators in excitatory and inhibitory cells for determining subtypes. The DNA methylation landscape of excitatory neurons in the cortex and hippocampus varied continuously along spatial gradients. Using this deep dataset, we constructed an artificial neural network model that precisely predicts single neuron cell-type identity and brain area spatial location. Integration of high-resolution DNA methylomes with single-nucleus chromatin accessibility data3 enabled prediction of high-confidence enhancer-gene interactions for all identified cell types, which were subsequently validated by cell-type-specific chromatin conformation capture experiments4. By combining multi-omic datasets (DNA methylation, chromatin contacts, and open chromatin) from single nuclei and annotating the regulatory genome of hundreds of cell types in the mouse brain, our DNA methylation atlas establishes the epigenetic basis for neuronal diversity and spatial organization throughout the mouse cerebrum.

An atlas of gene regulatory elements in adult mouse cerebrum.

The mammalian cerebrum performs high-level sensory perception, motor control and cognitive functions through highly specialized cortical and subcortical structures1. Recent surveys of mouse and human brains with single-cell transcriptomics2-6 and high-throughput imaging technologies7,8 have uncovered hundreds of neural cell types distributed in different brain regions, but the transcriptional regulatory programs that are responsible for the unique identity and function of each cell type remain unknown. Here we probe the accessible chromatin in more than 800,000 individual nuclei from 45 regions that span the adult mouse isocortex, olfactory bulb, hippocampus and cerebral nuclei, and use the resulting data to map the state of 491,818 candidate cis-regulatory DNA elements in 160 distinct cell types. We find high specificity of spatial distribution for not only excitatory neurons, but also most classes of inhibitory neurons and a subset of glial cell types. We characterize the gene regulatory sequences associated with the regional specificity within these cell types. We further link a considerable fraction of the cis-regulatory elements to putative target genes expressed in diverse cerebral cell types and predict transcriptional regulators that are involved in a broad spectrum of molecular and cellular pathways in different neuronal and glial cell populations. Our results provide a foundation for comprehensive analysis of gene regulatory programs of the mammalian brain and assist in the interpretation of noncoding risk variants associated with various neurological diseases and traits in humans.

DeepProg: an ensemble of deep-learning and machine-learning models for prognosis prediction using multi-omics data.

Multi-omics data are good resources for prognosis and survival prediction; however, these are difficult to integrate computationally. We introduce DeepProg, a novel ensemble framework of deep-learning and machine-learning approaches that robustly predicts patient survival subtypes using multi-omics data. It identifies two optimal survival subtypes in most cancers and yields significantly better risk-stratification than other multi-omics integration methods. DeepProg is highly predictive, exemplified by two liver cancer (C-index 0.73-0.80) and five breast cancer datasets (C-index 0.68-0.73). Pan-cancer analysis associates common genomic signatures in poor survival subtypes with extracellular matrix modeling, immune deregulation, and mitosis processes. DeepProg is freely available at https://github.com/lanagarmire/DeepProg.

Cardiac cell type-specific gene regulatory programs and disease risk association.

Misregulated gene expression in human hearts can result in cardiovascular diseases that are leading causes of mortality worldwide. However, the limited information on the genomic location of candidate cis-regulatory elements (cCREs) such as enhancers and promoters in distinct cardiac cell types has restricted the understanding of these diseases. Here, we defined >287,000 cCREs in the four chambers of the human heart at single-cell resolution, which revealed cCREs and candidate transcription factors associated with cardiac cell types in a region-dependent manner and during heart failure. We further found cardiovascular disease-associated genetic variants enriched within these cCREs including 38 candidate causal atrial fibrillation variants localized to cardiomyocyte cCREs. Additional functional studies revealed that two of these variants affect a cCRE controlling KCNH2/HERG expression and action potential repolarization. Overall, this atlas of human cardiac cCREs provides the foundation for illuminating cell type-specific gene regulation in human hearts during health and disease.

The MUC5B-associated variant rs35705950 resides within an enhancer subject to lineage- and disease-dependent epigenetic remodeling.

The G/T transversion rs35705950, located approximately 3 kb upstream of the MUC5B start site, is the cardinal risk factor for idiopathic pulmonary fibrosis (IPF). Here, we investigate the function and chromatin structure of this -3 kb region and provide evidence that it functions as a classically defined enhancer subject to epigenetic programming. We use nascent transcript analysis to show that RNA polymerase II loads within 10 bp of the G/T transversion site, definitively establishing enhancer function for the region. By integrating Assay for Transposase-Accessible Chromatin using sequencing (ATAC-seq) analysis of fresh and cultured human airway epithelial cells with nuclease sensitivity data, we demonstrate that this region is in accessible chromatin that affects the expression of MUC5B. Through applying paired single-nucleus RNA- and ATAC-seq to frozen tissue from IPF lungs, we extend these findings directly to disease, with results indicating that epigenetic programming of the -3 kb enhancer in IPF occurs in both MUC5B-expressing and nonexpressing lineages. In aggregate, our results indicate that the MUC5B-associated variant rs35705950 resides within an enhancer that is subject to epigenetic remodeling and contributes to pathologic misexpression in IPF.

Single-cell multiomic profiling of human lungs reveals cell-type-specific and age-dynamic control of SARS-CoV2 host genes.

Respiratory failure associated with COVID-19 has placed focus on the lungs. Here, we present single-nucleus accessible chromatin profiles of 90,980 nuclei and matched single-nucleus transcriptomes of 46,500 nuclei in non-diseased lungs from donors of ~30 weeks gestation,~3 years and ~30 years. We mapped candidate cis-regulatory elements (cCREs) and linked them to putative target genes. We identified distal cCREs with age-increased activity linked to SARS-CoV-2 host entry gene TMPRSS2 in alveolar type 2 cells, which had immune regulatory signatures and harbored variants associated with respiratory traits. At the 3p21.31 COVID-19 risk locus, a candidate variant overlapped a distal cCRE linked to SLC6A20, a gene expressed in alveolar cells and with known functional association with the SARS-CoV-2 receptor ACE2. Our findings provide insight into regulatory logic underlying genes implicated in COVID-19 in individual lung cell types across age. More broadly, these datasets will facilitate interpretation of risk loci for lung diseases.

Multimodal Meta-Analysis of 1,494 Hepatocellular Carcinoma Samples Reveals Significant Impact of Consensus Driver Genes on Phenotypes.

Although driver genes in hepatocellular carcinoma (HCC) have been investigated in various previous genetic studies, prevalence of key driver genes among heterogeneous populations is unknown. Moreover, the phenotypic associations of these driver genes are poorly understood. This report aims to reveal the phenotypic impacts of a group of consensus driver genes in HCC. We used MutSigCV and OncodriveFM modules implemented in the IntOGen pipeline to identify consensus driver genes across six HCC cohorts comprising 1,494 samples in total. To access their global impacts, we used The Cancer Genome Atlas (TCGA) mutations and copy-number variations to predict the transcriptomics data, under generalized linear models. We further investigated the associations of the consensus driver genes to patient survival, age, gender, race, and risk factors. We identify 10 consensus driver genes across six HCC cohorts in total. Integrative analysis of driver mutations, copy-number variations, and transcriptomic data reveals that these consensus driver mutations and their copy-number variations are associated with a majority (62.5%) of the mRNA transcriptome but only a small fraction (8.9%) of miRNAs. Genes associated with TP53, CTNNB1, and ARID1A mutations contribute to the tripod of most densely connected pathway clusters. These driver genes are significantly associated with patients' overall survival. Some driver genes are significantly linked to HCC gender (CTNNB1, ALB, TP53, and AXIN1), race (TP53 and CDKN2A), and age (RB1) disparities. This study prioritizes a group of consensus drivers in HCC, which collectively show vast impacts on the phenotypes. These driver genes may warrant as valuable therapeutic targets of HCC.

Deep Learning data integration for better risk stratification models of bladder cancer.

We propose an unsupervised multi-omics integration pipeline, using deep-learning autoencoder algorithm, to predict the survival subtypes in bladder cancer (BC). We used TCGA dataset comprising mRNA, miRNA and methylation to infer two survival subtypes. We then constructed a supervised classification model to predict the survival subgroups of any new individual sample. Our training data gave two subgroups with significant survival differences (p-value=8e-4), where high-risk survival subgroup was enriched with KRT6/14 overexpression and PI3K-Akt pathways. We tested the robustness of model by randomly splitting the main dataset into multiple training and test folds, which gave overall significant p-values. Then, we successfully inferred the subtypes for a subset of samples kept as test dataset (p-value=0.03). We further applied our pipeline to predict the survival subgroups from another validation dataset with miRNA data (p-value=0.02). Conclusively, present pipeline is an effective approach to infer the survival subtype of a new sample, exemplified by BC.

Deep Learning-Based Multi-Omics Integration Robustly Predicts Survival in Liver Cancer.

Identifying robust survival subgroups of hepatocellular carcinoma (HCC) will significantly improve patient care. Currently, endeavor of integrating multi-omics data to explicitly predict HCC survival from multiple patient cohorts is lacking. To fill this gap, we present a deep learning (DL)-based model on HCC that robustly differentiates survival subpopulations of patients in six cohorts. We built the DL-based, survival-sensitive model on 360 HCC patients' data using RNA sequencing (RNA-Seq), miRNA sequencing (miRNA-Seq), and methylation data from The Cancer Genome Atlas (TCGA), which predicts prognosis as good as an alternative model where genomics and clinical data are both considered. This DL-based model provides two optimal subgroups of patients with significant survival differences (P = 7.13e-6) and good model fitness [concordance index (C-index) = 0.68]. More aggressive subtype is associated with frequent TP53 inactivation mutations, higher expression of stemness markers (KRT19 and EPCAM) and tumor marker BIRC5, and activated Wnt and Akt signaling pathways. We validated this multi-omics model on five external datasets of various omics types: LIRI-JP cohort (n = 230, C-index = 0.75), NCI cohort (n = 221, C-index = 0.67), Chinese cohort (n = 166, C-index = 0.69), E-TABM-36 cohort (n = 40, C-index = 0.77), and Hawaiian cohort (n = 27, C-index = 0.82). This is the first study to employ DL to identify multi-omics features linked to the differential survival of patients with HCC. Given its robustness over multiple cohorts, we expect this workflow to be useful at predicting HCC prognosis prediction. Clin Cancer Res; 24(6); 1248-59. ©2017 AACR.

Single-Cell Transcriptomics Bioinformatics and Computational Challenges.

The emerging single-cell RNA-Seq (scRNA-Seq) technology holds the promise to revolutionize our understanding of diseases and associated biological processes at an unprecedented resolution. It opens the door to reveal intercellular heterogeneity and has been employed to a variety of applications, ranging from characterizing cancer cells subpopulations to elucidating tumor resistance mechanisms. Parallel to improving experimental protocols to deal with technological issues, deriving new analytical methods to interpret the complexity in scRNA-Seq data is just as challenging. Here, we review current state-of-the-art bioinformatics tools and methods for scRNA-Seq analysis, as well as addressing some critical analytical challenges that the field faces.