Search

Mammalian genomes possess multiple enhancers spanning an ultralong distance (>megabases) to modulate important genes, yet it is unclear how these enhancers coordinate to achieve this task. Here, we combine multiplexed CRISPRi screening with machine learning to define quantitative enhancer-enhancer interactions. We find that the ultralong distance enhancer network possesses a nested multi-layer architecture that confers functional robustness of gene expression. Experimental characterization reveals that enhancer epistasis is maintained by three-dimensional chromosomal interactions and BRD4 condensation. Machine learning prediction of synergistic enhancers provides an effective strategy to identify non-coding variant pairs associated with pathogenic genes in diseases beyond Genome-Wide Association Studies (GWAS) analysis. Our work unveils nested epistasis enhancer networks, which can better explain enhancer functions within cells and in diseases.

Systems genetics holds the promise to decipher complex traits by interpreting their associated SNPs through gene regulatory networks derived from comprehensive multi-omics data of cell types, tissues, and organs. Here, we propose SpecVar to integrate paired chromatin accessibility and gene expression data into context-specific regulatory network atlas and regulatory categories, conduct heritability enrichment analysis with genome-wide association studies (GWAS) summary statistics, identify relevant tissues, and estimate relevance correlation to depict common genetic factors acting in the shared regulatory networks between traits. Our method improves power upon existing approaches by associating SNPs with context-specific regulatory elements to assess heritability enrichments and by explicitly prioritizing gene regulations underlying relevant tissues. Ablation studies, independent data validation, and comparison experiments with existing methods on GWAS of six phenotypes show that SpecVar can improve heritability enrichment, accurately detect relevant tissues, and reveal causal regulations. Furthermore, SpecVar correlates the relevance patterns for pairs of phenotypes and better reveals shared SNP-associated regulations of phenotypes than existing methods. Studying GWAS of 206 phenotypes in UK Biobank demonstrates that SpecVar leverages the context-specific regulatory network atlas to prioritize phenotypes’ relevant tissues and shared heritability for biological and therapeutic insights. SpecVar provides a powerful way to interpret SNPs via context-specific regulatory networks and is available at https://github.com/AMSSwanglab/SpecVar, copy archived at swh:1:rev:cf27438d3f8245c34c357ec5f077528e6befe829.

Early cancer detection by cell-free DNA faces multiple challenges: low fraction of tumor cell-free DNA, molecular heterogeneity of cancer, and sample sizes that are not sufficient to reflect diverse patient populations. Here, we develop a cancer detection approach to address these challenges. It consists of an assay, cfMethyl-Seq, for cost-effective sequencing of the cell-free DNA methylome (with > 12-fold enrichment over whole genome bisulfite sequencing in CpG islands), and a computational method to extract methylation information and diagnose patients. Applying our approach to 408 colon, liver, lung, and stomach cancer patients and controls, at 97.9% specificity we achieve 80.7% and 74.5% sensitivity in detecting all-stage and early-stage cancer, and 89.1% and 85.0% accuracy for locating tissue-of-origin of all-stage and early-stage cancer, respectively. Our approach cost-effectively retains methylome profiles of cancer abnormalities, allowing us to learn new features and expand to other cancer types as training cohorts grow.

Elucidating the role of 3D architecture of DNA in gene regulation is crucial for understanding cell differentiation, tissue homeostasis and disease development. Among various chromatin conformation capture methods, HiChIP has received increasing attention for its significant improvement over other methods in profiling of regulatory (e.g. H3K27ac) and structural (e.g. cohesin) interactions. To facilitate the studies of 3D regulatory interactions, we developed a HiChIP interactions database, HiChIPdb (http://health.tsinghua.edu.cn/hichipdb/). The current version of HiChIPdb contains ∼262M annotated HiChIP interactions from 200 high-throughput HiChIP samples across 108 cell types. The functionalities of HiChIPdb include: (i) standardized categorization of HiChIP interactions in a hierarchical structure based on organ, tissue and cell line and (ii) comprehensive annotations of HiChIP interactions with regulatory genes and GWAS Catalog SNPs. To the best of our knowledge, HiChIPdb is the first comprehensive database that utilizes a unified pipeline to map the functional interactions across diverse cell types and tissues in different resolutions. We believe this database has the potential to advance cutting-edge research in regulatory mechanisms in development and disease by removing the barrier in data aggregation, preprocessing, and analysis.

Non-coding sequences that are evolutionarily conserved and bio-chemically active offer clues to mechanistic interpretations of human genome-wide association studies (GWAS). However, their genetic effects have not been systematically examined across a wide range of human tissues and traits. Here we develop a simple method to identify functional elements exhibiting high levels of human-mouse sequence conservation and enhancer-like biochemical activity, which scales well to 313 epigenomic datasets across 106 tissues and cell types. Combining these elements with 468 GWAS of European (EUR) and East Asian (EAS) ancestries, we identify tissue-specific enrichments of heritability and causal variants for many traits, as well as candidate genes that are functionally relevant to body mass index (BMI) and schizophrenia but were not reported in previous GWAS. Our findings provide a comprehensive assessment of how sequence-conserved enhancer-like elements affect complex traits, and reinforce the importance of integrating evolutionary and biochemical data to elucidate human disease genetics.

Background The 15q13.3 microdeletion is associated with several neuropsychiatric disorders, including autism and schizophrenia. Previous association and functional studies have investigated the potential role of several genes within the deletion in neuronal dysfunction, but the molecular effects of the deletion as a whole remain largely unknown. Methods Induced pluripotent stem cells, from 3 patients with the 15q13.3 microdeletion and 3 control subjects, were generated and converted into induced neurons. We analyzed the effects of the 15q13.3 microdeletion on genome-wide gene expression, DNA methylation, chromatin accessibility, and sensitivity to cisplatin-induced DNA damage. Furthermore, we measured gene expression changes in induced neurons with CRISPR (clustered regularly interspaced short palindromic repeats) knockouts of individual 15q13.3 microdeletion genes. Results In both induced pluripotent stem cells and induced neurons, gene copy number change within the 15q13.3 microdeletion was accompanied by significantly decreased gene expression and no compensatory changes in DNA methylation or chromatin accessibility, supporting the model that haploinsufficiency of genes within the deleted region drives the disorder. Furthermore, we observed global effects of the microdeletion on the transcriptome and epigenome, with disruptions in several neuropsychiatric disorder–associated pathways and gene families, including Wnt signaling, ribosome function, DNA binding, and clustered protocadherins. Individual gene knockouts mirrored many of the observed changes in an overlapping fashion between knockouts. Conclusions Our multiomics analysis of the 15q13.3 microdeletion revealed downstream effects in pathways previously associated with neuropsychiatric disorders and indications of interactions between genes within the deletion. This molecular systems analysis can be applied to other chromosomal aberrations to further our etiological understanding of neuropsychiatric disorders.

Technological development has enabled the profiling of gene expression and chromatin accessibility from the same cell. We develop scREG, a dimension reduction methodology, based on the concept of cis-regulatory potential, for single cell multiome data. This concept is further used for the construction of subpopulation-specific cis-regulatory networks. The capability of inferring useful regulatory network is demonstrated by the two-fold increment on network inference accuracy compared to the Pearson correlation-based method and the 27-fold enrichment of GWAS variants for inflammatory bowel disease in the cis-regulatory elements. The R package scREG provides comprehensive functions for single cell multiome data analysis.

Deciphering the free energy landscape of biomolecular structure space is crucial for understanding many complex molecular processes, such as protein-protein interaction, RNA folding, and protein folding. A major source of current dynamic structure data is Molecular Dynamics (MD) simulations. Several methods have been proposed to investigate the free energy landscape from MD data, but all of them rely on the assumption that kinetic similarity is associated with global geometric similarity, which may lead to unsatisfactory results. In this paper, we proposed a new method called Conditional Angle Partition Tree to reveal the hierarchical free energy landscape by correlating local geometric similarity with kinetic similarity. Its application on the benchmark alanine dipeptide MD data showed a much better performance than existing methods in exploring and understanding the free energy landscape. We also applied it to the MD data of Villin HP35. Our results are more reasonable on various aspects than those from other methods and very informative on the hierarchical structure of its energy landscape.

Significance Here we use the expression and accessibility data from a diverse set of cell types to learn a model for the dependence of the accessibility of a regulatory element on its DNA sequence and TF expression. Using GTEx samples with WGS data, we show that the noncoding variants predicted to affect accessibility are more strongly associated with the expression of nearby genes. To interpret a personal genome, we combine the sequence information with context-specific TF expression to prioritize variants and regulatory elements in any genomic region of interest. This approach should be helpful in the study of risk loci previously identified by GWAS. Results from analysis of height and WGS data from the GTEx project support this hypothesis., A person’s genome typically contains millions of variants which represent the differences between this personal genome and the reference human genome. The interpretation of these variants, i.e., the assessment of their potential impact on a person’s phenotype, is currently of great interest in human genetics and medicine. We have developed a prioritization tool called OpenCausal which takes as inputs 1) a personal genome and 2) a reference context-specific TF expression profile and returns a list of noncoding variants prioritized according to their impact on chromatin accessibility for any given genomic region of interest. We applied OpenCausal to 6,430 samples across 18 tissues derived from the GTEx project and found that the variants prioritized by OpenCausal are highly enriched for eQTLs and caQTLs. We further propose a strategy to integrate the predicted open scores with genome-wide association studies (GWAS) data to prioritize putative causal variants and regulatory elements for a given risk locus (i.e., fine-mapping analysis). As an initial example, we applied this method to a GWAS dataset of human height and found that the prioritized putative variants and elements are correlated with the phenotype (i.e., heights of individuals) better than others.

Shotgun assays are widely used in biotechnologies to characterize large molecules, which are hard to be measured as a whole directly. For instance, in Liquid Chromatography - Mass Spectrometry (LC-MS) shotgun experiments, proteins in biological samples are digested into peptides, and then peptides are separated and measured. However, in proteomics study, investigators are usually interested in the performance of the whole proteins instead of those peptide fragments. In light of meta-analysis, we propose an adaptive thresholding method to select informative peptides, and combine peptide-level models to protein-level analysis. The meta-analysis procedure and modeling rationale can be adapted to data analysis of other types of shotgun assays.

The genetic information coded in DNA leads to trait innovation via a gene regulatory network (GRN) in development. Here, we developed a conserved non-coding element interpretation method to integrate multi-omics data into gene regulatory network (CNEReg) to investigate the ruminant multi-chambered stomach innovation. We generated paired expression and chromatin accessibility data during rumen and esophagus development in sheep and revealed 1601 active ruminant-specific conserved non-coding elements (active-RSCNEs). To interpret the function of these active-RSCNEs, we define toolkit transcription factors (TTFs) and model their regulation on rumen-specific genes via batteries of active-RSCNEs during development. Our developmental GRN revealed 18 TTFs and 313 active-RSCNEs regulating seven rumen functional modules. Notably, six TTFs (OTX1, SOX21, HOXC8, SOX2, TP63, and PPARG), as well as 16 active-RSCNEs, functionally distinguish the rumen from the esophagus. Our study provides a systematic approach to understanding how gene regulation evolves and shapes complex traits by putting evo-devo concepts into practice with developmental multi-omics data.

Significance Abdominal aortic aneurysm (AAA) is a common and severe disease with major genetic risk factors. In this study we generated enhancer-promoter contact data to identify regulatory elements in AAA-relevant cell types and identified changes in their predicted chromatin accessibility between AAA patients and controls. We integrated this information with disease-associated variants in regulatory elements and gene bodies to further understand the etiology and pathogenetic mechanisms of AAA. Our study combined whole-genome sequencing data with gene regulatory relations in disease-relevant cell types to reveal the important roles of the interleukin 6 pathway and ERG and KLF regulation in AAA pathogenesis., Abdominal aortic aneurysm (AAA) is a common degenerative cardiovascular disease whose pathobiology is not clearly understood. The cellular heterogeneity and cell-type-specific gene regulation of vascular cells in human AAA have not been well-characterized. Here, we performed analysis of whole-genome sequencing data in AAA patients versus controls with the aim of detecting disease-associated variants that may affect gene regulation in human aortic smooth muscle cells (AoSMC) and human aortic endothelial cells (HAEC), two cell types of high relevance to AAA disease. To support this analysis, we generated H3K27ac HiChIP data for these cell types and inferred cell-type-specific gene regulatory networks. We observed that AAA-associated variants were most enriched in regulatory regions in AoSMC, compared with HAEC and CD4+ cells. The cell-type-specific regulation defined by this HiChIP data supported the importance of ERG and the KLF family of transcription factors in AAA disease. The analysis of regulatory elements that contain noncoding variants and also are differentially open between AAA patients and controls revealed the significance of the interleukin-6-mediated signaling pathway. This finding was further validated by including information from the deleteriousness effect of nonsynonymous single-nucleotide variants in AAA patients and additional control data from the Medical Genome Reference Bank dataset. These results shed important insights into AAA pathogenesis and provide a model for cell-type-specific analysis of disease-associated variants.

BackgroundBiologists long recognized that the genetic information encoded in DNA leads to trait innovation via gene regulatory network (GRN) in development.ResultsHere, we generated paired expression and chromatin accessibility data during rumen and esophagus development in sheep and revealed 1,601 active ruminant-specific conserved non-coding elements (active-RSCNEs). To interpret the function of these active-RSCNEs, we developed a Conserved Non-coding Element interpretation method by gene Regulatory network (CNEReg) to define toolkit transcription factors (TTF) and model its regulation on rumen specific gene via batteries of active-RSCNEs during development. Our developmental GRN reveals 18 TTFs and 313 active-RSCNEs regulating the functional modules of the rumen and identifies OTX1, SOX21, HOXC8, SOX2, TP63, PPARG and 16 active-RSCNEs that functionally distinguish the rumen from the esophagus.ConclusionsWe argue that CNEReg is an attractive systematic approach to integrate evo-devo concepts with omics data to understand how gene regulation evolves and shapes complex traits.

Standard methods for the creation of neuronal cells via direct induction from primary tissue use perinatal fibroblasts, which hinders the important study of patient specific genetic lesions such as those underlying neuropsychiatric disorders. To address this we developed a novel method for the direct induction of neuronal cells (induced neuronal cells, iN cells) from adult human fibroblast cells. Reprogramming fibroblasts into iN cells via recombinant virus resulted in cells that stain for markers such as MAP2 and PSA-NCAM and exhibit electrophysiological properties such as action potentials and voltage dependent sodium- and potassium currents that reveal a neuronal phenotype. Transcriptome and chromatin analysis using RNA-Seq, microRNA-Seq and ATAC-Seq, respectively, further confirm neuronal character. 22q11.2 Deletion-Syndrome (22q11DS) is caused by a large 3 million base-pair heterozygous deletion on human chromosome 22 and is strongly associated with neurodevelopmental, neuropsychiatric phenotypes such as schizophrenia and autism. We leverage the direct-iN cell model for the study of genetic neurodevelopmental conditions by presenting gene-by-gene as well as networkwide effects of the 22q11DS deletion on gene expression in human neuronal cells, on several levels of functional genomics analysis. Some of the genes within the 22q11DS deletion boundary exhibit unexpected cell-type-specific changes in transcript levels, and genome-wide we can detect dysregulation of calcium channel subunit genes and other genes known to be involved in autism or schizophrenia, such as NRXN1, as well synaptic pathways. This genome-wide effect on gene expression can also be observed at the microRNA and chromatin levels, showing that the iN cells have indeed converted to a neuronal phenotype at several regulatory levels: chromatin, protein-coding RNAs and microRNAs, revealing relevant disease pathways and genes. We present this model of inducing neurons from fibroblasts as a useful general resource to study the genetic and molecular basis of normal and abnormal brain development and brain function.

Deciphering the free energy landscape of biomolecular structure space is crucial for understanding many complex molecular processes, such as protein-protein interaction, RNA folding, and protein folding. A major source of current dynamic structure data is Molecular Dynamics (MD) simulations. Several methods have been proposed to investigate the free energy landscape from MD data, but all of them rely on the assumption that kinetic similarity is associated with global geometric similarity, which may lead to unsatisfactory results. In this paper, we proposed a new method called Conditional Angle Partition Tree to reveal the hierarchical free energy landscape by correlating local geometric similarity with kinetic similarity. Its application on the benchmark alanine dipeptide MD data showed a much better performance than existing methods in exploring and understanding the free energy landscape. We also applied it to the MD data of Villin HP35. Our results are more reasonable on various aspects than those from other methods and very informative on the hierarchical structure of its energy landscape.

The comparison of gene regulatory networks between diseased versus healthy individuals or between two different treatments is an important scientific problem. Here, we propose sc-compReg as a method for the comparative analysis of gene expression regulatory networks between two conditions using single cell gene expression (scRNA-seq) and single cell chromatin accessibility data (scATAC-seq). Our software, sc-compReg, can be used as a stand-alone package that provides joint clustering and embedding of the cells from both scRNA-seq and scATAC-seq, and the construction of differential regulatory networks across two conditions. We apply the method to compare the gene regulatory networks of an individual with chronic lymphocytic leukemia (CLL) versus a healthy control. The analysis reveals a tumor-specific B cell subpopulation in the CLL patient and identifies TOX2 as a potential regulator of this subpopulation., Changes in cell state underlie the difference between health and disease. Here, the authors propose a computational framework for the integration of gene expression and chromatin-accessibility data from single cells to identify differences in gene regulation in cell types across two conditions.

Head pose estimation has attracted immense research interest recently, as its inherent information significantly improves the performance of face-related applications such as face alignment and face recognition. In this paper, we conduct an in-depth study of head pose estimation and present a multiregression loss function, an $L2$ regression loss combined with an ordinal regression loss, to train a convolutional neural network (CNN) that is dedicated to estimating head poses from RGB images without depth information. The ordinal regression loss is utilized to address the nonstationary property observed as the facial features change with respect to different head pose angles and learn robust features. The $L2$ regression loss leverages these features to provide precise angle predictions for input images. To avoid the ambiguity problem in the commonly used Euler angle representation, we further formulate the head pose estimation problem in quaternions. Our quaternion-based multiregression loss method achieves state-of-the-art performance on the AFLW2000, AFLW test set, and AFW datasets and is closing the gap with methods that utilize depth information on the BIWI dataset.

Complex traits such as cardiovascular diseases (CVD) are the results of complicated processes jointly affected by genetic and environmental factors. Genome-wide association studies (GWAS) identified genetic variants associated with diseases but usually did not reveal the underlying mechanisms. There could be many intermediate steps at epigenetic, transcriptomic, and cellular scales inside the black box of genotype-phenotype associations. In this article, we present a machine-learning-based cross-scale framework GRPath to decipher putative causal paths (pcPaths) from genetic variants to disease phenotypes by integrating multiple omics data. Applying GRPath on CVD, we identified 646 and 549 pcPaths linking putative causal regions, variants, and gene expressions in specific cell types for two types of heart failure, respectively. The findings suggest new understandings of coronary heart disease. Our work promoted the modeling of tissue- and cell type-specific cross-scale regulation to uncover mechanisms behind disease-associated variants, and provided new findings on the molecular mechanisms of CVD.

Bayesian Sequential Partitioning (BSP) is a statistically effective density estimation method to comprehend the characteristics of a high dimensional data space. The intensive computation of the statistical model and the counting of enormous data have caused serious design challenges for BSP to handle the growing volume of the data. This paper proposes a high performance design of BSP by leveraging a heterogeneous CPU/GPGPU system that consists of a host CPU and a K80 GPGPU. A series of techniques, on both data structures and execution management policies, is implemented to extensively exploit the computation capability of the heterogeneous many-core system and alleviate system bottlenecks. When compared with a parallel design on a high-end CPU, the proposed techniques achieve 48x average runtime enhancement while the maximum speedup can reach 78.76x.

Significance Density estimation is among the most fundamental problems in statistics. It is notoriously difficult to estimate the density of high-dimensional data due to the “curse of dimensionality.” Here, we introduce a new general-purpose density estimator based on deep generative neural networks. By modeling data normally distributed around a manifold of reduced dimension, we show how the power of bidirectional generative neural networks (e.g., cycleGAN) can be exploited for explicit evaluation of the data density. Simulation and real data experiments suggest that our method is effective in a wide range of problems. This approach should be helpful in many applications where an accurate density estimator is needed., Density estimation is one of the fundamental problems in both statistics and machine learning. In this study, we propose Roundtrip, a computational framework for general-purpose density estimation based on deep generative neural networks. Roundtrip retains the generative power of deep generative models, such as generative adversarial networks (GANs) while it also provides estimates of density values, thus supporting both data generation and density estimation. Unlike previous neural density estimators that put stringent conditions on the transformation from the latent space to the data space, Roundtrip enables the use of much more general mappings where target density is modeled by learning a manifold induced from a base density (e.g., Gaussian distribution). Roundtrip provides a statistical framework for GAN models where an explicit evaluation of density values is feasible. In numerical experiments, Roundtrip exceeds state-of-the-art performance in a diverse range of density estimation tasks.

Discovering DNA regulatory sequence motifs and their relative positions are vital to understand the mechanisms of gene expression regulation. Such complicated motif grammars are difficult to be summarized from shallow models. Although Deep Convolutional Neural Network (DCNN) achieved great success in annotating cis-regulatory elements, few combinatorial motif grammars have been accurately interpreted due to the mixed signal in DCNN. To address this problem, we proposed NeuronMotif, a general backward decoupling algorithm, to reveal the homo-/hetero-typic motif combinations and arrangements embedded in convolutional neurons. We applied NeuronMotif on several widely-used DCNN models. Many uncovered motif grammars of deep convolutional neurons are supported by literature or ATAC-seq footprinting. We further diagnosed the sick neurons that are sensitive to adversarial noises, which can guide DCNN architecture optimization for better prediction performance and motif feature extraction. Overall, NeuronMotif enables decoding cis-regulatory codes from deep convolutional neurons and understanding DCNN from a novel perspective.

Although computational approaches have been complementing high-throughput biological experiments for the identification of functional regions in the human genome, it remains a great challenge to systematically decipher interactions between transcription factors (TFs) and regulatory elements to achieve interpretable annotations of chromatin accessibility across diverse cellular contexts. To solve this problem, we propose DeepCAGE, a deep learning framework that integrates sequence information and binding statuses of TFs, for the accurate prediction of chromatin accessible regions at a genome-wide scale in a variety of cell types. DeepCAGE takes advantage of a densely connected deep convolutional neural network architecture to automatically learn sequence signatures of known chromatin accessible regions and then incorporates such features with expression levels and binding activities of human core TFs to predict novel chromatin accessible regions. In a series of systematic comparisons with existing methods, DeepCAGE exhibits superior performance in not only the classification but also the regression of chromatin accessibility signals. In a detailed analysis of TF activities, DeepCAGE successfully extracts novel binding motifs and measures the contribution of a TF to the regulation with respect to a specific locus in a certain cell type. When applied to whole-genome sequencing data analysis, our method successfully prioritizes putative deleterious variants underlying a human complex trait and thus provides insights into the understanding of disease-associated genetic variants. DeepCAGE can be downloaded from https://github.com/kimmo1019/DeepCAGE.

Single-cell multi-omics data continues to grow at an unprecedented pace, and effectively integrating different modalities holds the promise for better characterization of cell identities. Although a number of methods have demonstrated promising results in integrating multiple modalities from the same tissue, the complexity and scale of data compositions typically present in cell atlases still pose a significant challenge for existing methods. Here we present scJoint, a transfer learning method to integrate atlas-scale, heterogeneous collections of scRNA-seq and scATAC-seq data. scJoint leverages information from annotated scRNA-seq data in a semi-supervised framework and uses a neural network to simultaneously train labeled and unlabeled data, enabling label transfer and joint visualization in an integrative framework. Using multiple atlas data and a biologically varying multi-modal data, we demonstrate scJoint is computationally efficient and consistently achieves significantly higher cell type label accuracy than existing methods while providing meaningful joint visualizations. This suggests scJoint is effective in overcoming the heterogeneity in different modalities towards a more comprehensive understanding of cellular phenotypes.

In multilabel classification, strong label dependence is present for exploiting, particularly for word-to-word dependence defined by semantic labels. In such a situation, we develop a collaborative-learning framework to predict class labels based on label-predictor pairs and label-only data. For example, in image categorization and recognition, language expressions describe the content of an image together with a large number of words and phrases without associated images. This article proposes a new loss quantifying partial correctness for false positive and negative misclassifications due to label similarities. Given this loss, we develop the Bayes rule to capture label dependence by nonlinear classification. On this ground, we introduce a weighted random forest classifier for complete data and a stacking scheme for leveraging additional labels to enhance the performance of supervised learning based on label-predictor pairs. Importantly, we decompose multilabel classification into a sequence of independent learning tasks, based on which the computational complexity of our classifier becomes linear in the size of labels. Compared to existing classifiers without label-only data, the proposed classifier enjoys the computational benefit while enabling the detection of novel labels absent from training by exploring label dependence and leveraging label-only data for higher accuracy. Theoretically, we show that the proposed method reconstructs the Bayes performance consistently, achieving the desired learning accuracy. Numerically, we demonstrate that the proposed method compares favorably in terms of the proposed and Hamming losses against binary relevance and a regularized Ising classifier modeling conditional label dependence. Indeed, leveraging additional labels tends to improve the supervised performance, especially when the training sample is not very large, as in semisupervised learning. Finally, we demonstrate the utility of the proposed approach on the Microsoft COCO object detection challenge, PASCAL visual object classes challenge 2007, and Mediamill benchmark.

Multi-omics data allow us to select a small set of informative markers for the discrimination of specific cell types and study of cellular heterogeneity. However, it is often challenging to choose an optimal marker panel from the high-dimensional molecular profiles for a large amount of cell types. Here, we propose a method called Mixed Integer programming Model to Identify Cell type-specific marker panel (MIMIC). MIMIC maintains the hierarchical topology among different cell types and simultaneously maximizes the specificity of a fixed number of selected markers. MIMIC was benchmarked on the mouse ENCODE RNA-seq dataset, with 29 diverse tissues, for 43 surface markers (SMs) and 1345 transcription factors (TFs). MIMIC could select biologically meaningful markers and is robust for different accuracy criteria. It shows advantages over the standard single gene-based approaches and widely used dimensional reduction methods, such as multidimensional scaling and t-SNE, both in accuracy and in biological interpretation. Furthermore, the combination of SMs and TFs achieves better specificity than SMs or TFs alone. Applying MIMIC to a large collection of 641 RNA-seq samples covering 231 cell types identifies a panel of TFs and SMs that reveal the modularity of cell type association networks. Finally, the scalability of MIMIC is demonstrated by selecting enhancer markers from mouse ENCODE data. MIMIC is freely available at https://github.com/MengZou1/MIMIC.

Recent advances in single-cell technologies, including single-cell ATAC-seq (scATAC-seq), have enabled large-scale profiling of the chromatin accessibility landscape at the single cell level. However, the characteristics of scATAC-seq data, including high sparsity and high dimensionality, have greatly complicated the computational analysis. Here, we proposed scDEC, a computational tool for single cell ATAC-seq analysis with deep generative neural networks. scDEC is built on a pair of generative adversarial networks (GANs), and is capable of learning the latent representation and inferring the cell labels, simultaneously. In a series of experiments, scDEC demonstrates superior performance over other tools in scATAC-seq analysis across multiple datasets and experimental settings. In the downstream applications, we demonstrated that the generative power of scDEC helps to infer the trajectory and intermediate state of cells during differentiation and the latent features learned by scDEC can potentially reveal both biological cell types and within-cell-type variations.

Unsupervised methods, including clustering methods, are essential to the analysis of single-cell genomic data. Model-based clustering methods are under-explored in the area of single-cell genomics, and have the advantage of quantifying the uncertainty of the clustering result. Here we develop a model-based approach for the integrative analysis of single-cell chromatin accessibility and gene expression data. We show that combining these two types of data, we can achieve a better separation of the underlying cell types. An efficient Markov chain Monte Carlo algorithm is also developed.

We have developed a generally applicable method based on CRISPR/Cas9-targeted ultra-long read sequencing (CTLR-Seq) to completely and haplotype-specifically resolve, at base-pair resolution, large, complex, and highly repetitive genomic regions that had been previously impenetrable to next-generation sequencing analysis such as large segmental duplication (SegDup) regions and their associated genome rearrangements that stretch hundreds of kilobases. Our method combines in vitro Cas9-mediated cutting of the genome and pulse-field gel electrophoresis to haplotype-specifically isolate intact large (200-550 kb) target regions that encompass previously unresolvable genomic sequences. These target fragments are then sequenced (amplification-free) to produce ultra-long reads at up to 40x on-target coverage using Oxford nanopore technology, allowing for the complete assembly of the complex genomic regions of interest at single base-pair resolution. We applied CTLR-Seq to resolve the exact sequence of SegDup rearrangements that constitute the boundary regions of the 22q11.2 deletion CNV and of the 16p11.2 deletion and duplication CNVs. These CNVs are among the strongest known risk factors for schizophrenia and autism. We then perform de novo assembly to resolve, for the first time, at single base-pair resolution, the sequence rearrangements of the 22q11.2 and 16p11.2 CNVs, mapping out exactly the genes and non-coding regions that are affected by the CNV for different carriers.

Significance Biological samples are often heterogeneous mixtures of different types of cells. Suppose we have two single-cell datasets, each providing information on a different cellular feature and generated on a different sample from this mixture. Then, the clustering of cells in the two samples should be coupled as both clusterings are reflecting the underlying cell types in the same mixture. This “coupled clustering” problem is a new problem not covered by existing clustering methods. In this paper, we develop an approach for its solution based on the coupling of two nonnegative matrix factorizations. The method should be useful for integrative single-cell genomics analysis tasks such as the joint analysis of single-cell RNA-sequencing and single-cell ATAC-sequencing data., When different types of functional genomics data are generated on single cells from different samples of cells from the same heterogeneous population, the clustering of cells in the different samples should be coupled. We formulate this “coupled clustering” problem as an optimization problem and propose the method of coupled nonnegative matrix factorizations (coupled NMF) for its solution. The method is illustrated by the integrative analysis of single-cell RNA-sequencing (RNA-seq) and single-cell ATAC-sequencing (ATAC-seq) data.

Although genome-wide association studies (GWAS) have successfully identified thousands of genomic loci associated with hundreds of complex traits in the past decade, the debate about such problems as missing heritability and weak interpretability has been appealing for effective computational methods to facilitate the advanced analysis of the vast volume of existing and anticipated genetic data. Towards this goal, gene-level integrative GWAS analysis with the assumption that genes associated with a phenotype tend to be enriched in biological gene sets or gene networks has recently attracted much attention, due to such advantages as straightforward interpretation, less multiple testing burdens, and robustness across studies. However, existing methods in this category usually exploit non-tissue-specific gene networks and thus lack the ability to utilize informative tissue-specific characteristics. To overcome this limitation, we proposed a Bayesian approach called SIGNET (Simultaneously Inference of GeNEs and Tissues) to integrate GWAS data and multiple tissue-specific gene networks for the simultaneous inference of phenotype-associated genes and relevant tissues. Through extensive simulation studies, we showed the effectiveness of our method in finding both associated genes and relevant tissues for a phenotype. In applications to real GWAS data of 14 complex phenotypes, we demonstrated the power of our method in both deciphering genetic basis and discovering biological insights of a phenotype. With this understanding, we expect to see SIGNET as a valuable tool for integrative GWAS analysis, thereby boosting the prevention, diagnosis, and treatment of human inherited diseases and eventually facilitating precision medicine.

Early cancer detection by cell-free DNA (cfDNA) faces multiple challenges: the low fraction of tumor DNA in cfDNA, the molecular heterogeneity of cancer, and sample sizes that are too small to reflect the heterogeneous patient population. We have developed an integrated cancer detection system, CancerRadar, that addresses all three challenges. It consists of (1) a cost-effective experimental assay, cfMethyl-Seq, for genome-wide methylation profiling of cfDNA, which provides >12-fold enrichment over Whole Genome Bisulfite Sequencing (WGBS) in CpG islands; and (2) a computational platform to extract information from cfMethyl-Seq data and diagnose the patient. The platform derives cfDNA methylations, cfDNA fragment sizes, copy number variations (CNV), and microbial composition from the raw cfMethyl-Seq data, and performs multi-feature ensemble learning. We demonstrate the power of CancerRadar by detecting and locating cancer in a cohort of 275 colon, liver, lung, and stomach cancer patients and 204 non-cancer individuals. For cancer detection, we achieve a sensitivity of 85.6%± 6.7% across all stages and 80.6%±9.1% for early stages (I and II), with a specificity of 99% in both cases. These metrics are derived using leave-one-out cross-validation. During independent validation on a reserved subsample, it achieves a sensitivity of 89.1%±11.3% across all stages and 85.7%±14.2% for early stages, with a specificity of 97% (one false positive). For locating a tumor's tissue of origin (TOO), CancerRadar achieved an accuracy of 91.5%±5.0% for all stages and 89.1%±7.3% for early stages, on an independent subsample. This study is the first to integrate cfDNA methylation, cfDNA fragment size, CNV, and microbial composition analyses for cancer detection on the same patient cohort. cfDNA methylation was the most useful for detecting cancer, but including features from other categories significantly increased the performance, especially for early-stage cancer. In contrast, with respect to TOO prediction, methylation-derived features were overwhelmingly important while including other features did not further improve performance. To fully exploit the power of cfDNA methylation, we identified four types of methylation markers with different characteristics. We have also improved our previous read-level deconvolution algorithm to more accurately identify trace tumor signals. Finally, our data show that as training sample sizes increase, the detection power of CancerRadar continues to increase. Although all existing cancer detection studies are limited by training sample sizes, the CancerRadar system uniquely and cost-effectively retains the genome-wide epigenetic and genetic profiles of cancer abnormalities, thereby permitting the classification models to learn and exploit newly significant features as training cohorts grow, as well as expanding their scope to other cancer types. Citation Format: Mary Stackpole, Weihua Zeng, Shuo Li, Chun-Chi Liu, Yonggang Zhou, Shanshan He, Angela Yeh, Ziye Wang, Fengzhu Sun, Qingjiao Li, Zuyang Yuan, Asli Yildirim, Pin Jung Chen, Paul Winograd, Shize Li, Zorawar Noor, Edward Garon, Samuel French, Clara Magyar, Sarah Dry, Clara Lajonchere, Daniel Geschwind, Gina Choi, Sammy Saab, Frank Alber, Wing Hung Wong, Steven Dubinett, Denise Aberle, Vatche Agopian, Steven-Huy Han, Xiaohui Ni, Wenyuan Li, Xianghong Jasmine Zhou. Multi-feature ensemble learning on cell-free dna for accurately detecting and locating cancer [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2021; 2021 Apr 10-15 and May 17-21. Philadelphia (PA): AACR; Cancer Res 2021;81(13_Suppl):Abstract nr 24.

An energy-efficient restricted Boltzmann machine (RBM) processor (RBM-P) supporting on-chip learning and inference is proposed for machine learning and Internet of Things (IoT) applications in this paper. To train a neural network (NN) model, the RBM structure is applied to supervised and unsupervised learning, and a multi-layer NN can be constructed and initialized by stacking multiple RBMs. Featuring NN model reduction for external memory bandwidth saving, low power neuron binarizer (LPNB) with dynamic clock gating and area-efficient NN-like activation function calculators for power reduction, user-defined connection map (UDCM) for both computation time and bandwidth saving, and early stopping (ES) mechanism for learning process, the proposed system integrates 32 RBM cores with maximal 4k neurons per layer and 128 candidates per sample for machine learning applications. Implemented in 65nm CMOS technology, the proposed RBM-P chip costs 2.2 M gates and 128 kB SRAM with 8.8 mm2 area. Operated at 1.2 V and 210 MHz, this chip achieves 7.53G neuron weights (NWs) and 11.63G NWs per second with 41.3 and 26.7 pJ per NW for learning and inference, respectively.

Although computational approaches have been complementing high-throughput biological experiments for the identification of functional regions in the human genome, it remains a great challenge to systematically decipher interactions between transcription factors and regulatory elements to achieve interpretable annotations of chromatin accessibility across diverse cellular contexts. Towards this problem, we propose DeepCAGE, a deep learning framework that integrates sequence information and binding status of transcription factors, for the accurate prediction of chromatin accessible regions at a genome-wide scale in a variety of cell types. DeepCAGE takes advantage of a densely connected deep convolutional neural network architecture to automatically learn sequence signatures of known chromatin accessible regions, and then incorporates such features with expression levels and binding activities of human core transcription factors to predict novel chromatin accessible regions. In a series of systematic comparisons with existing methods, DeepCAGE exhibits superior performance in not only the classification but also the regression of chromatin accessibility signals. In detailed analysis of transcription factor activities, DeepCAGE successfully extracts novel binding motifs and measures the contribution of a transcription factor to the regulation with respect to a specific locus in a certain cell type. When applied to whole-genome sequencing data analysis, our method successfully prioritizes putative deleterious variants underlying a human complex trait, and thus provides insights into the understanding of disease-associated genetic variants. DeepCAGE can be downloaded from https://github.com/kimmo1019/DeepCAGE.

Characterizing and interpreting heterogeneous mixtures at the cellular level is a critical problem in genomics. Single-cell assays offer an opportunity to resolve cellular level heterogeneity, e.g., scRNA-seq enables single-cell expression profiling, and scATAC-seq identifies active regulatory elements. Furthermore, while scHi-C can measure the chromatin contacts (i.e., loops) between active regulatory elements to target genes in single cells, bulk HiChIP can measure such contacts in a higher resolution. In this work, we introduce DC3 (De-Convolution and Coupled-Clustering) as a method for the joint analysis of various bulk and single-cell data such as HiChIP, RNA-seq and ATAC-seq from the same heterogeneous cell population. DC3 can simultaneously identify distinct subpopulations, assign single cells to the subpopulations (i.e., clustering) and de-convolve the bulk data into subpopulation-specific data. The subpopulation-specific profiles of gene expression, chromatin accessibility and enhancer-promoter contact obtained by DC3 provide a comprehensive characterization of the gene regulatory system in each subpopulation., Single-cell omics analysis can reveal heterogeneity among individual cells at different levels. Here, the authors develop DC3, a computational method for joint analysis of various bulk and single-cell data from the same heterogeneous cell population.

Heterozygous deletions in the 15q13.3 region are associated with several neuropsychiatric disorders including autism, schizophrenia, and attention deficit hyperactivity disorder. Several genes within the 15q13.3 deletion region may play a role in neuronal dysfunction, based on association studies in humans and functional studies in mice, but the intermediate molecular mechanisms remain unknown. We analyzed the genome-wide effects of the 15q13.3 microdeletion on the transcriptome and epigenome. Induced pluripotent stem cell (iPSC) lines from three patients with the typical heterozygous 15q13.3 microdeletion and three sex-matched controls were generated and converted into induced neurons (iNs) using the neurogenin-2 induction method. We analyzed genome-wide gene expression using RNA-Seq, genome-wide DNA methylation using SeqCap-Epi, and genome-wide chromatin accessibility using ATAC-Seq, in both iPSCs and iNs. In both cell types, gene copy number change within the 15q13.3 microdeletion was accompanied by significantly decreased gene expression and no compensatory changes in DNA methylation or chromatin accessibility, supporting the model that haploinsufficiency of genes within the deleted region drives the disorder. Further, we observed global effects of the deletion on the transcriptome and epigenome, with the effects being cell type specific and occurring at discrete loci. Several genes and pathways associated with neuropsychiatric disorders and neuronal development were significantly altered, including Wnt signaling, ribosome biogenesis, DNA binding, and clustered protocadherins. This molecular systems analysis of a large neuropsychiatric microdeletion can also be applied to other brain relevant chromosomal aberrations to further our etiological understanding of neuropsychiatric disorders.

We study a class of non-parametric density estimators under Bayesian settings. The estimators are obtained by adaptively partitioning the sample space. Under a suitable prior, we analyze the concentration rate of the posterior distribution, and demonstrate that the rate does not directly depend on the dimension of the problem in several special cases. Another advantage of this class of Bayesian density estimators is that it can adapt to the unknown smoothness of the true density function, thus achieving the optimal convergence rate without artificial conditions on the density. We also validate the theoretical results on a variety of simulated data sets.

Cell packs a lot of genetic and regulatory information through a structure known as chromatin, i.e. DNA is wrapped around histone proteins and is tightly packed in a remarkable way. To express a gene in a specific coding region, the chromatin would open up and DNA loop may be formed by interacting enhancers and promoters. Furthermore, the mediator and cohesion complexes, sequence-specific transcription factors, and RNA polymerase II are recruited and work together to elaborately regulate the expression level. It is in pressing need to understand how the information, about when, where, and to what degree genes should be expressed, is embedded into chromatin structure and gene regulatory elements. Thanks to large consortia such as Encyclopedia of DNA Elements (ENCODE) and Roadmap Epigenomic projects, extensive data on chromatin accessibility and transcript abundance are available across many tissues and cell types. This rich data offer an exciting opportunity to model the causal regulatory relationship. Here, we will review the current experimental approaches, foundational data, computational problems, interpretive frameworks, and integrative models that will enable the accurate interpretation of regulatory landscape. Particularly, we will discuss the efforts to organize, analyze, model, and integrate the DNA accessibility data, transcriptional data, and functional genomic regions together. We believe that these efforts will eventually help us understand the information flow within the cell and will influence research directions across many fields.

Liquid biopsy using cell-free DNA (cfDNA) is attractive for a wide range of clinical applications, including cancer detection, locating, and monitoring. However, developing these applications requires precise and sensitive calling of somatic single nucleotide variations (SNVs) from cfDNA sequencing data. To date, no SNV caller comprehensively addresses the unique challenges of cfDNA to provide reliable results. Here we present cfSNV, a revolutionary somatic SNV caller with five innovative techniques to overcome and exploit the unique properties of cfDNA: (1) a joint-genotype statistical model, to modeling the mixed nature of cfDNA and thus probabilistically deconvolute the mixture of tumor- and normal-derived reads; (2) hierarchical mutation profiling to detect clusters of mutations with different variant allele frequencies; (3) a data pre-processing step that exploits the overlapping read mates caused by short cfDNA fragments to improve data quality; (4) a new filtering procedure to remove misalignment errors that account for the non-random fragmentation pattern of cfDNA; and (5) a machine learning approach that incorporates the sequencing context to filter errors at the level of individual reads. These five techniques permit cfSNV to far outperform existing methods in sensitively and reliably detecting mutations in cfDNA, even with medium-coverage WES data. On both simulated data and real patient data, we demonstrate that cfSNV is superior to existing tools, especially for low-frequency somatic SNVs. On simulated data, without sacrificing precision, cfSNV showed 3.0- and 3.7-fold increase in sensitivity of all somatic SNVs, and specifically 11.1- and 138.5-fold increase in sensitivity of somatic SNVs with allele frequency < 1% comparing to MuTect and Strelka2 respectively. The superior performance of cfSNV has been further confirmed on real patient data from 18 metastatic breast cancer and prostate cancer patients. Besides comparison with existing tools, we also showed how the five novel techniques contribute to its performance. The applicability of cfSNV to WES data enables cfDNA to be used in a wide variety of clinical applications, and we present one example of such an application: a novel and effective immunotherapy response measure that we name truncal-bTMB. This measure exploits cfDNA's ability to provide comprehensive coverage of the mutation landscape and hence describe the clonal structure. We demonstrated this new marker allows a better selection of non-small-cell lung cancer patients for immunotherapy treatment. Citation Format: Shuo Li, Zorawar Noor, Weihua Zeng, Xiaohui Ni, Zuyang Yuan, Frank Alber, Wenyuan Li, Edward B. Garon, Xianghong Zhou, Wing Hung Wong, Vatche G. Agopian, Steven M. Dubinett M. Dubinett. Sensitive detection of tumor mutations from blood and its application to immunotherapy prognosis [abstract]. In: Proceedings of the Annual Meeting of the American Association for Cancer Research 2020; 2020 Apr 27-28 and Jun 22-24. Philadelphia (PA): AACR; Cancer Res 2020;80(16 Suppl):Abstract nr LB-247.

High-resolution interactions among regulatory elements are of crucial importance for the understanding of transcriptional regulation and the interpretation of disease mechanism. Hi-C technique allows the genome-wide detection of chromatin contacts. However, unless extremely deep sequencing is performed on a very large number of input cells, current Hi-C experiments do not have high enough resolution to resolve contacts among regulatory elements. Here, we develop DeepTACT, a bootstrapping deep learning model, to integrate genome sequences and chromatin accessibility data for the prediction of chromatin contacts among regulatory elements. In tests based on promoter capture Hi-C data, DeepTACT is seen to offer improved resolution over existing methods. DeepTACT analysis also identifies a class of hub promoters, which are active across cell lines, enriched in housekeeping genes, functionally related to fundamental biological processes, and capable of reflecting cell similarity. Finally, the utility of high-resolution chromatin contact information in the study of human diseases is illustrated by the association of IFNA2 and IFNA1 to coronary artery disease via an integrative analysis of GWAS data and high-resolution contacts inferred by DeepTACT.

Characterizing epigenetic heterogeneity at the cellular level is a critical problem in the modern genomics era. Assays such as single cell ATAC-seq (scATAC-seq) offer an opportunity to interrogate cellular level epigenetic heterogeneity through patterns of variability in open chromatin. However, these assays exhibit technical variability that complicates clear classification and cell type identification in heterogeneous populations. We present scABC, an R package for the unsupervised clustering of single-cell epigenetic data, to classify scATAC-seq data and discover regions of open chromatin specific to cell identity., Single cell ATAC-seq (scATAC-seq) data reveals cellular level epigenetic heterogeneity but its application in delineating distinct subpopulations is still challenging. Here, the authors develop scABC, a statistical method for unsupervised clustering of scATAC-seq data and identification of open chromatin regions specific to cell identity.

SUMMARYTissue development results from lineage-specific transcription factors (TF) programming a dynamic chromatin landscape through progressive cell fate transitions. Here, we interrogate the epigenomic landscape during epidermal differentiation and create an inference network that ranks the coordinate effects of TF-accessible regulatory element-target gene expression triplets on lineage commitment. We discover two critical transition periods: surface ectoderm initiation and keratinocyte maturation, and identify TFAP2C and p63 as lineage initiation and maturation factors, respectively. Surprisingly, we find that TFAP2C, and not p63, is sufficient to initiate surface ectoderm differentiation, with TFAP2C-initiated progenitor cells capable of maturing into functional keratinocytes. Mechanistically, TFAP2C primes the surface ectoderm chromatin landscape and induces p63 expression and binding sites, thus allowing maturation factor p63 to positively auto-regulate its expression and close a subset of the TFAP2C-initiated early program. Our work provides a general framework to infer TF networks controlling chromatin transitions that will facilitate future regenerative medicine advances.

When different types of functional genomics data are generated on single cells from different samples of cells from the same heterogeneous population, the clustering of cells in the different samples should be coupled. We formulate this “coupled clustering” problem as an optimization problem, and propose the method of coupled nonnegative matrix factorizations (coupled NMF) for its solution. The method is illustrated by the integrative analysis of single cell RNA-seq and single cell ATAC-seq data.Significance StatementsBiological samples are often heterogeneous mixtures of different types of cells. Suppose we have two single cell data sets, each providing information on a different cellular feature and generated on a different sample from this mixture. Then, the clustering of cells in the two samples should be coupled as both clusterings are reflecting the underlying cell types in the same mixture. This “coupled clustering” problem is a new problem not covered by existing clustering methods. In this paper we develop an approach for its solution based the coupling of two nonnegative matrix factorizations. The method should be useful for integrative single cell genomics analysis tasks such as the joint analysis of single cell RNA-seq and single cell ATAC-seq data.

Finding the locations and identities of faces in videos is a very important task in numerous applications. In this paper, we propose a correlation-based face detection approach to improve the performance of face recognition tasks for videos. We apply correlation measures to pairs of response maps which are generated from automatically selected neurons in deep convolutional neural network (CNN) models to detect faces in each video frame. The embeddings extracted from faces cropped by our proposed approach are more consistent across each video sequence and more suitable for face recognition and clustering tasks. Experimental results from the YouTube Faces (YTF) dataset demonstrate that our proposed approach is more robust and achieves better recognition accuracy compared to state-of-the-art face detection approaches.

Diabetic retinopathy is the primary cause of blindness in the working-age population of the developed world. Diagnosing the disease heavily relies on imaging studies, which is a time consuming and a manual process performed by trained clinicians. Enhancing the accuracy and speed of the detection process can potentially have a significant impact on population health via early diagnosis and intervention. Motivated by this, we propose a recognition pipeline based on deep convolutional neural networks. In our pipeline, we design lightweight networks called SI2DRNet-vl along with six methods to further boost the detection performance. Without any fine-tuning, our recognition pipeline outperforms state of the art on the Messidor dataset along with 5.26x fewer in total parameters and 2.48x fewer in total floating operations.

In this paper, we propose Confidence Network (ConfNet) which not only makes predictions on input images but also generates a confidence score that estimates the probability of correctness of each prediction. Furthermore, Confidence Loss is proposed to make ConfNet automatically learn confidence scores in the training phase. The experiments on two public datasets show that the confidence scores generated by ConfNet are highly correlated with the model accuracy and outperforms two related methods. When stacking two ConfNets in a cascade structure, 3.8x computational cost can be saved compared to the single state-of-the-art model with only 0.1 % increase of error rate.

This paper presents an implementation of an energy efficient bit-plane payload design for machine learning processor. The proposed architecture facilitates high parallelism and high data bandwidth and thus improves the model learning/training time of machine learning algorithms. By assembling multiple bits as a bit-plane and enlarging query parallelism with a central compare-flag updater, data processing parallelism can be increased. Binary sequential partition (BSP), a fast density estimation algorithm capable of dealing with high dimensional data sets, is realized. Fabricated in 90nm 1P9M CMOS process, the processing rate can achieve 16.9 Gb/sec with 8 queries for data dimension D=210. The test chip integrates 64 counting cells and provides 5 modes with power consumptions of 1.86mJ/Gb per Query.

Approximate Bayesian Computation (ABC) methods are used to approximate posterior distributions in models with unknown or computationally intractable likelihoods. Both the accuracy and computational efficiency of ABC depend on the choice of summary statistic, but outside of special cases where the optimal summary statistics are known, it is unclear which guiding principles can be used to construct effective summary statistics. In this paper we explore the possibility of automating the process of constructing summary statistics by training deep neural networks to predict the parameters from artificially generated data: the resulting summary statistics are approximately posterior means of the parameters. With minimal model-specific tuning, our method constructs summary statistics for the Ising model and the moving-average model, which match or exceed theoretically-motivated summary statistics in terms of the accuracies of the resulting posteriors., 27 pages, 10 figures