Search

Motivation In silico saturation mutagenesis (ISM) is a popular approach in computational genomics for calculating feature attributions on biological sequences that proceeds by systematically perturbing each position in a sequence and recording the difference in model output. However, this method can be slow because systematically perturbing each position requires performing a number of forward passes proportional to the length of the sequence being examined. Results In this work, we propose a modification of ISM that leverages the principles of compressed sensing to require only a constant number of forward passes, regardless of sequence length, when applied to models that contain operations with a limited receptive field, such as convolutions. Our method, named Yuzu, can reduce the time that ISM spends in convolution operations by several orders of magnitude and, consequently, Yuzu can speed up ISM on several commonly used architectures in genomics by over an order of magnitude. Notably, we found that Yuzu provides speedups that increase with the complexity of the convolution operation and the length of the sequence being analyzed, suggesting that Yuzu provides large benefits in realistic settings. Availability and implementation We have made this tool available at https://github.com/kundajelab/yuzu.

High-throughput experimental platforms have revolutionized the ability to profile biochemical and functional properties of biological sequences such as DNA, RNA and proteins. By collating several data modalities with customizable tracks rendered using intuitive visualizations, genome browsers enable an interactive and interpretable exploration of diverse types of genome profiling experiments and derived annotations. However, existing genome browser tracks are not well suited for intuitive visualization of high-resolution DNA sequence features such as transcription factor motifs. Typically, motif instances in regulatory DNA sequences are visualized as BED-based annotation tracks, which highlight the genomic coordinates of the motif instances but do not expose their specific sequences. Instead, a genome sequence track needs to be cross-referenced with the BED track to identify sequences of motif hits. Even so, quantitative information about the motif instances such as affinity or conservation as well as differences in base resolution from the consensus motif are not immediately apparent. This makes interpretation slow and challenging. This problem is compounded when analyzing several cellular states and/or molecular readouts (such as ATAC-seq and ChIP–seq) simultaneously, as coordinates of enriched regions (peaks) and the set of active transcription factor motifs vary across cell states.

Inbred mouse strains reveal the molecular basis of mammalian traits and diseases, particularly recessive ones. We utilized mouse community curated resources to set up an automated screen to discover novel testable gene function hypotheses. Using 11,832 community contributed strain-differentiating experiments and trait presence/absence scoring, we searched for all experiments where strains can be split by their phenotypic values (e.g., high vs. low responders). Then, using 48 sequenced strains, we found one or more candidate gene for each experiment where homozygous high-impact variants (such as stopgain, frameshifts) segregate strains into these same binary grouping. Our approach rediscovered 212 known gene-phenotype relationships, almost always highlighting potentially novel causal variants, as well as thousands of gene function hypotheses. To help find the most exciting hypotheses, we improved the state of the art in machine learning driven literature-based discovery (LBD). Reading on our top 3 ranked candidate genes per experiment reveals 80% of rediscovered relationships, compared to 5% reading at random. We proposed 1,842 novel gene-phenotype testable hypotheses using our approach. We built a web portal at aimhigh.stanford.edu to allow researchers to view all our testable hypotheses in detail. Our open-source code can be rerun as more sequenced strains and phenotyping experiments become available.

We introduce the dynseq genome browser track, which displays DNA nucleotide characters scaled by user-specified, base-resolution scores provided in the BigWig file format. The dynseq track enables visualization of context-specific, informative genomic sequence features. We demonstrate its utility in three popular genome browsers for interpreting cis-regulatory sequence syntax and regulatory variant interpretation by visualizing nucleotide importance scores derived from machine learning models of regulatory DNA trained on protein-DNA binding and chromatin accessibility experiments.

Background: Smooth muscle cells (SMCs) transition into a number of different phenotypes during atherosclerosis, including those that resemble fibroblasts and chondrocytes, and make up the majority of cells in the atherosclerotic plaque. To better understand the epigenetic and transcriptional mechanisms that mediate these cell state changes, and how they relate to risk for coronary artery disease (CAD), we have investigated the causality and function of transcription factors at genome-wide associated loci. Methods: We used CRISPR-Cas 9 genome and epigenome editing to identify the causal gene and cells for a complex CAD genome-wide association study signal at 2q22.3. Single-cell epigenetic and transcriptomic profiling in murine models and human coronary artery smooth muscle cells were used to understand the cellular and molecular mechanism by which this CAD risk gene exerts its function. Results: CRISPR-Cas 9 genome and epigenome editing showed that the complex CAD genetic signals within a genomic region at 2q22.3 lie within smooth muscle long-distance enhancers for ZEB2 , a transcription factor extensively studied in the context of epithelial mesenchymal transition in development of cancer. Zeb2 regulates SMC phenotypic transition through chromatin remodeling that obviates accessibility and disrupts both Notch and transforming growth factor β signaling, thus altering the epigenetic trajectory of SMC transitions. SMC-specific loss of Zeb2 resulted in an inability of transitioning SMCs to turn off contractile programing and take on a fibroblast-like phenotype, but accelerated the formation of chondromyocytes, mirroring features of high-risk atherosclerotic plaques in human coronary arteries. Conclusions: These studies identify ZEB2 as a new CAD genome-wide association study gene that affects features of plaque vulnerability through direct effects on the epigenome, providing a new therapeutic approach to target vascular disease.

Data availability: No new data were generated or analysed in support of this research. Copyright © The Author(s) 2022. GENCODE produces high quality gene and transcript annotation for the human and mouse genomes. All GENCODE annotation is supported by experimental data and serves as a reference for genome biology and clinical genomics. The GENCODE consortium generates targeted experimental data, develops bioinformatic tools and carries out analyses that, along with externally produced data and methods, support the identification and annotation of transcript structures and the determination of their function. Here, we present an update on the annotation of human and mouse genes, including developments in the tools, data, analyses and major collaborations which underpin this progress. For example, we report the creation of a set of non-canonical ORFs identified in GENCODE transcripts, the LRGASP collaboration to assess the use of long transcriptomic data to build transcript models, the progress in collaborations with RefSeq and UniProt to increase convergence in the annotation of human and mouse protein-coding genes, the propagation of GENCODE across the human pan-genome and the development of new tools to support annotation of regulatory features by GENCODE. Our annotation is accessible via Ensembl, the UCSC Genome Browser and https://www.gencodegenes.org. National Human Genome Research Institute of the National Institutes of Health [U41HG007234, R01HG004037]; Wellcome Trust [WT222155/Z/20/Z]; European Molecular Biology Laboratory. The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Institutes of Health. Funding for open access charge: National Institutes of Health.

SummaryCongenital heart defects, the most common birth disorders, are the clinical manifestation of anomalies in fetal heart development - a complex process involving dynamic spatiotemporal coordination among various precursor cell lineages. This complexity underlies the incomplete understanding of the genetic architecture of congenital heart diseases (CHDs). To define the multi-cellular epigenomic and transcriptional landscape of cardiac cellular development, we generated single-cell chromatin accessibility maps of human fetal heart tissues. We identified eight major differentiation trajectories involving primary cardiac cell types, each associated with dynamic transcription factor (TF) activity signatures. We identified similarities and differences of regulatory landscapes of iPSC-derived cardiac cell types and their in vivo counterparts. We interpreted deep learning models that predict cell-type resolved, base-resolution chromatin accessibility profiles from DNA sequence to decipher underlying TF motif lexicons and infer the regulatory impact of non-coding variants. De novo mutations predicted to affect chromatin accessibility in arterial endothelium were enriched in CHD cases versus controls. We used CRISPR-based perturbations to validate an enhancer harboring a nominated regulatory CHD mutation, linking it to effects on the expression of a known CHD gene JARID2. Together, this work defines the cell-type resolved cis-regulatory sequence determinants of heart development and identifies disruption of cell type-specific regulatory elements as a component of the genetic etiology of CHD.

Genome-wide association studies (GWAS) of eye disorders have identified hundreds of genetic variants associated with ocular disease. However, the vast majority of these variants are noncoding, making it challenging to interpret their function. Here, we present a joint single-cell atlas of gene expression and chromatin accessibility of the adult human retina with >50,000 cells, which we used to analyze noncoding single-nucleotide polymorphisms (SNPs) implicated by GWAS of age-related macular degeneration, glaucoma, diabetic retinopathy, myopia, and type 2 macular telangiectasia. We integrate this atlas with a HiChIP enhancer connectome, expression quantitative trait loci (eQTL) data, and base-resolution deep learning models to predict noncoding SNPs with causal roles in eye disease, assess SNP impact on transcription factor binding, and define their known and novel target genes. Our efforts nominate pathogenic SNP-target gene interactions for multiple vision disorders and provide a potentially powerful resource for interpreting noncoding variation in the eye.

Somatic cell transcription factors are critical to maintaining cellular identity and constitute a barrier to human somatic cell reprogramming; yet a comprehensive understanding of the mechanism of action is lacking. To gain insight, we examined epigenome remodeling at the onset of human nuclear reprogramming by profiling human fibroblasts after fusion with murine embryonic stem cells (ESCs). By assay for transposase-accessible chromatin with high-throughput sequencing (ATAC-seq) and chromatin immunoprecipitation sequencing we identified enrichment for the activator protein 1 (AP-1) transcription factor c-Jun at regions of early transient accessibility at fibroblast-specific enhancers. Expression of a dominant negative AP-1 mutant (dnAP-1) reduced accessibility and expression of fibroblast genes, overcoming the barrier to reprogramming. Remarkably, efficient reprogramming of human fibroblasts to induced pluripotent stem cells was achieved by transduction with vectors expressing SOX2, KLF4, and inducible dnAP-1, demonstrating that dnAP-1 can substitute for exogenous human OCT4. Mechanistically, we show that the AP-1 component c-Jun has two unexpected temporally distinct functions in human reprogramming: 1) to potentiate fibroblast enhancer accessibility and fibroblast-specific gene expression, and 2) to bind to and repress OCT4 as a complex with MBD3. Our findings highlight AP-1 as a previously unrecognized potent dual gatekeeper of the somatic cell state.

Aging is accompanied by a deterioration in the regenerative and repair potential of stem cell niches in the brain1–5. However, the mechanisms underlying this decline are largely unknown. Here we characterize genome-wide chromatin accessibility in young and old neurogenic niche cellsin vivo, revealing defects in neural stem cell (NSC) adhesion and migration during aging. Interestingly, chromatin accessibility at cell adhesion and migration genes decreases with age in quiescent NSCs but increases with age in activated (proliferative) NSCs, and this is accompanied by corresponding expression changes in these genes. We experimentally validate that quiescent and activated NSCs exhibit opposing adhesion and migration behaviors with age: quiescent NSCs become less adhesive (and more migratory) whereas activated NSCs and progeny become more adhesive (and less migratory) during aging. We also show that the ability of activated NSCs and progeny to mobilize out of the niche duringin vivoneurogenesis diminishes during aging. Using tension sensors with single molecule resolution, we find that one of the cellular mechanisms by which aging impairs the migration of old activated NSCs and progeny involves increased force-producing adhesions. We identify inhibition of the cytoskeletal-regulating kinase ROCK6, 7as a way to reduce force-producing adhesions and restore migration in old activated NSCsin vitro. Interestingly, inhibition of ROCK in the neurogenic niche of old mice boosts neurogenesis to the olfactory bulbin vivo. These results have important implications for restoring the migratory potential of NSCs and progeny and for improving neurogenesis in the aged brain.

Deep learning models such as convolutional neural networks are able to accurately map biological sequences to associated functional readouts and properties by learning predictive de novo representations. In-silico saturation mutagenesis (ISM) is a popular feature attribution technique for inferring contributions of all characters in an input sequence to the model’s predicted output. The main drawback of ISM is its runtime, as it involves multiple forward propagations of all possible mutations of each character in the input sequence through the trained model to predict the effects on the output. We present fastISM, an algorithm that speeds up ISM by a factor of over 10x for commonly used convolutional neural network architectures. fastISM is based on the observations that the majority of computation in ISM is spent in convolutional layers, and a single mutation only disrupts a limited region of intermediate layers, rendering most computation redundant. fastISM reduces the gap between backpropagation-based feature attribution methods and ISM. It far surpasses the runtime of backpropagation-based methods on multi-output architectures, making it feasible to run ISM on a large number of sequences. An easy-to-use Keras/TensorFlow 2 implementation of fastISM is available at https://github.com/kundajelab/fastISM, and a hands-on tutorial at https://colab.research.google.com/github/kundajelab/fastISM/blob/master/notebooks/colab/DeepSEA.ipynb.

Motivation Genome-wide profiles of chromatin accessibility and gene expression in diverse cellular contexts are critical to decipher the dynamics of transcriptional regulation. Recently, convolutional neural networks have been used to learn predictive cis-regulatory DNA sequence models of context-specific chromatin accessibility landscapes. However, these context-specific regulatory sequence models cannot generalize predictions across cell types. Results We introduce multi-modal, residual neural network architectures that integrate cis-regulatory sequence and context-specific expression of trans-regulators to predict genome-wide chromatin accessibility profiles across cellular contexts. We show that the average accessibility of a genomic region across training contexts can be a surprisingly powerful predictor. We leverage this feature and employ novel strategies for training models to enhance genome-wide prediction of shared and context-specific chromatin accessible sites across cell types. We interpret the models to reveal insights into cis- and trans-regulation of chromatin dynamics across 123 diverse cellular contexts. Availability and implementation The code is available at https://github.com/kundajelab/ChromDragoNN. Supplementary information Supplementary data are available at Bioinformatics online.

The AlphaZero approach got a great success and achieved superhuman performance across many challenging games, but we think there are at least three problems that can be improved. Firstly, AlphaZero only estimates win, draw, or lose but ignores how many points it will get or lose. Secondly, AlphaZero uses Monte-Carlo Tree Search to derive an average value among all the children nodes' values in a subtree. Thirdly, AlphaZero does not consider the depth rewards during the Monte-Carlo Tree Search. To solve these three problems, we introduce a general-purpose framework, the Big-Best-Quick win strategy in Monte-Carlo Tree Search, to try to surpass the AlphaZero approach. In this paper, we mainly focus on the Big-win strategy to improve the performance of AlphaZero without human knowledge. We are pleased to derive some promising results in which our Big-win approach has improved the strength of the 6x6 Othello program with win rate=63%, lose rate=28%, and draw rate=9% comparing to the original AlphaZero approach based on a fair training and playing time conditions.

The relationship between noncoding DNA sequence and gene expression is not well-understood. Massively parallel reporter assays (MPRAs), which quantify the regulatory activity of large libraries of DNA sequences in parallel, are a powerful approach to characterize this relationship. We present MPRA-DragoNN, a convolutional neural network (CNN)-based framework to predict and interpret the regulatory activity of DNA sequences as measured by MPRAs. While our method is generally applicable to a variety of MPRA designs, here we trained our model on the Sharpr-MPRA dataset that measures the activity of ~500,000 constructs tiling 15,720 regulatory regions in human K562 and HepG2 cell lines. MPRA-DragoNN predictions were moderately correlated (Spearmanρ= 0.28) with measured activity and were within range of replicate concordance of the assay. State-of-the-art model interpretation methods revealed high-resolution predictive regulatory sequence features that overlapped transcription factor (TF) binding motifs. We used the model to investigate the cell type and chromatin state preferences of predictive TF motifs. We explored the ability of our model to predict the allelic effects of regulatory variants in an independent MPRA experiment and fine map putative functional SNPs in loci associated with lipid traits. Our results suggest that interpretable deep learning models trained on MPRA data have the potential to reveal meaningful patterns in regulatory DNA sequences and prioritize regulatory genetic variants, especially as larger, higher-quality datasets are produced.

We define the novel problem of extracting and predicting occurrence dates for a class of recurrent events -- events that are held periodically as per a near-regular schedule (e.g., conferences, film festivals, sport championships). Knowledge-bases such as Freebase contain a large number of such recurring events, but they also miss substantial information regarding specific event instances and their occurrence dates. We develop a temporal extraction and inference engine to fill in the missing dates as well as to predict their future occurrences. Our engine performs joint inference over several knowledge sources -- (1) information about an event instance and its date extracted from text by our temporal extractor, (2) information about the typical schedule (e.g., ``every second week of June") for a recurrent event extracted by our schedule extractor, and (3) known dates for other instances of the same event. The output of our system is a representation for the event schedule and an occurrence date for each event instance. We find that our system beats humans in predicting future occurrences of recurrent events by significant margins. We release our code and system output for further research.