A5070578596- Single-cell and spatial transcriptomics
- Genomics and Chromatin Dynamics
- Chemical Reactions and Isotopes
- Epigenetics and DNA Methylation
- Radiopharmaceutical Chemistry and Applications
- Genomics and Phylogenetic Studies
- Gene Regulatory Network Analysis
- RNA and protein synthesis mechanisms
- Gene expression and cancer classification
- Machine Learning in Bioinformatics
- Bioinformatics and Genomic Networks
- Topic Modeling
- Microtubule and mitosis dynamics
- RNA Research and Splicing
- Glioma Diagnosis and Treatment
- RNA modifications and cancer
- Extracellular vesicles in disease
- MicroRNA in disease regulation
- CRISPR and Genetic Engineering
- Algorithms and Data Compression
- Cell Image Analysis Techniques
- Machine Learning in Healthcare
- Domain Adaptation and Few-Shot Learning
- Biomedical Text Mining and Ontologies
- Cancer Research and Treatment
Brown University
2020-2025
John Brown University
2020-2025
Seattle University
2019-2020
University of Washington
2019-2020
University of Virginia
2014-2018
Office of Public Health Genomics
2015
Hertfordshire Community NHS Trust
2008
Histone modifications are among the most important factors that control gene regulation. Computational methods predict expression from histone modification signals highly desirable for understanding their combinatorial effects in This knowledge can help developing 'epigenetic drugs' diseases like cancer. Previous studies quantifying relationship between and levels either failed to capture or relied on multiple separate predictions analysis. paper develops a unified discriminative framework...
Abstract Imaging chromatin dynamics is crucial to understand genome organization and its role in transcriptional regulation. Recently, the RNA-guidable feature of CRISPR-Cas9 has been utilized for imaging within live cells. However, these methods are mostly applicable highly repetitive regions, whereas regions with low or no repeats remains as a challenge. To address this challenge, we design single-guide RNAs (sgRNAs) integrated up 16 MS2 binding motifs enable robust fluorescent signal...
The CRISPR system has become a powerful biological tool with wide range of applications. However, improving targeting specificity and accurately predicting potential off-targets remains significant goal. Here, we introduce web-based CRISPR/Cas9 Off-target Prediction Identification Tool (CROP-IT) that performs improved off-target binding cleavage site predictions. Unlike existing prediction programs solely use DNA sequence information; CROP-IT integrates whole genome level information from...
Recent advances in sequencing technologies have allowed us to capture various aspects of the genome at single-cell resolution. However, with exception a few co-assaying technologies, it is not possible simultaneously apply different assays on same single cell. In this scenario, computational integration multi-omic measurements crucial enable joint analyses. This task particularly challenging due lack sample-wise or feature-wise correspondences. We present alignment optimal transport (SCOT),...
Abstract Several age predictors based on DNA methylation, dubbed epigenetic clocks, have been created in recent years, with the vast majority regularized linear regression. This study explores improvement performance and interpretation of clocks using deep learning. First, we gathered 142 publicly available data sets from several human tissues to develop AltumAge, a neural network framework that is highly accurate precise predictor. Compared ElasticNet, AltumAge performs better for...
The MLL gene is a common target of chromosomal translocations found in human leukemia. MLL-fusion leukemia has consistently poor outcome. One the most translocation partners AF9 (MLLT3). MLL-AF9 recruits DOT1L, histone 3 lysine 79 methyltransferase (H3K79me1/me2/me3), leading to aberrant transcription. We show that DOT1L three binding sites and present nuclear magnetic resonance (NMR) solution structure DOT1L-AF9 complex. generate structure-guided point mutations find they have graded...
Genes or their encoded products are not expected to mingle with each other unless in some disease situations. In cancer, a frequent mechanism that can produce gene fusions is chromosomal rearrangement. However, recent discoveries of RNA trans-splicing and cis-splicing between adjacent genes (cis-SAGe) support for mechanisms generating fusion RNAs. our transcriptome analyses 28 prostate normal cancer samples, 30% RNAs on average the transcripts contain exons belonging same-strand neighboring...
Abstract Many single-cell sequencing technologies are now available, but it is still difficult to apply multiple the same single cell. In this paper, we propose an unsupervised manifold alignment algorithm, MMD-MA, for integrating measurements carried out on disjoint aliquots of a given population cells. Effectively, MMD-MA performs in silico co-assay by embedding cells measured different ways into learned latent space. data points from domains aligned optimizing objective function with...
Abstract Objective Alzheimer’s disease (AD) is the most common neurodegenerative disorder with one of complex pathogeneses, making effective and clinically actionable decision support difficult. The objective this study was to develop a novel multimodal deep learning framework aid medical professionals in AD diagnosis. Materials Methods We present Multimodal Disease Diagnosis (MADDi) accurately detect presence mild cognitive impairment (MCI) from imaging, genetic, clinical data. MADDi that...
Chitinase 3-like 1 (Chi3l1) is a secreted protein that highly expressed in glioblastoma. Here, we show Chi3l1 alters the state of glioma stem cells (GSC) to support tumor growth. Exposure patient-derived GSCs reduced frequency CD133+SOX2+ and increased CD44+Chi3l1+ cells. bound CD44 induced phosphorylation nuclear translocation β-catenin, Akt, STAT3. Single-cell RNA sequencing velocity following incubation with showed significant changes GSC dynamics driving towards mesenchymal expression...
Patients with rare and complex diseases often experience delayed diagnoses misdiagnoses because comprehensive knowledge about these is limited to only a few medical experts. In this context, large language models (LLMs) have emerged as powerful aggregation tools applications in clinical decision support education domains. This study aims explore the potential of 3 popular LLMs, namely Bard (Google LLC), ChatGPT-3.5 (OpenAI), GPT-4 enhance diagnosis while investigating impact prompt...
Aging is a complex and multifaceted process involving many epigenetic alterations. One key area of interest in aging research the role histone modifications, which can dynamically regulate gene expression. Here, we conducted pan-tissue analysis dynamics seven modifications during human aging. Our histone-specific age prediction models showed surprisingly accurate performance, proving resilient to experimental artificial noise. Simulation experiments for comparison with DNA methylation...
Abstract Data integration of single-cell measurements is critical for understanding cell development and disease, but the lack correspondence between different types makes such efforts challenging. Several unsupervised algorithms can align heterogeneous in a shared space, enabling creation mappings single cells data domains. However, these require hyperparameter tuning high-quality alignments, which difficult an setting without information validation. We present Single-Cell alignment using...
While COVID-19 diagnosis and prognosis artificial intelligence models exist, very few can be implemented for practical use given their high risk of bias. We aimed to develop a model that addresses notable shortcomings prior studies, integrating it into fully automated triage pipeline examines chest radiographs the presence, severity, progression pneumonia. Scans were collected using DICOM Image Analysis Archive, system communicates with hospital's image repository. The authors over 6,500...
This paper applies a deep convolutional/highway MLP framework to classify genomic sequences on the transcription factor binding site task. To make model understandable, we propose an optimization driven strategy extract "motifs", or symbolic patterns which visualize positive class learned by network. We show that our system, Deep Motif (DeMo), extracts motifs are similar to, and in some cases outperform current well known motifs. In addition, find deeper consisting of multiple convolutional...
Integrating single-cell measurements that capture different properties of the genome is vital to extending our understanding biology. This task challenging due lack a shared axis across datasets obtained from types experiments. For most such datasets, we corresponding information among cells (samples) and (features). In this scenario, unsupervised algorithms are capable aligning experiments critical learning an in silico co-assay can help draw correspondences cells. Maximum mean...
Abstract Machine learning models that predict genomic activity are most useful when they make accurate predictions across cell types. Here, we show the training and test sets contain same loci, resulting model may falsely appear to perform well by effectively memorizing average associated with each locus We demonstrate this phenomenon in context of predicting gene expression chromatin domain boundaries, suggest methods diagnose avoid pitfall. anticipate that, as more data becomes available,...
The three-dimensional organization of genomes plays a crucial role in essential biological processes. segregation chromatin into A and B compartments highlights regions activity inactivity, providing window the genomic activities specific to each cell type. Yet, steep costs associated with acquiring Hi-C data, necessary for studying this compartmentalization across various types, pose significant barrier type genome organization. To address this, we present prediction tool called compartment...
Abstract Summary Measurement of single-cell gene expression at different timepoints enables the study cell development. However, due to resource constraints and technical challenges associated with experiments, researchers can only profile discrete sparsely sampled timepoints. This missing timepoint information impedes downstream developmental analyses. We propose scNODE, an end-to-end deep learning model that predict in silico unobserved scNODE integrates a variational autoencoder neural...
Long-range regulatory interactions among genomic regions are critical for controlling gene expression, and their disruption has been associated with a host of diseases. However, when modeling the effects factors, most deep learning models either neglect long-range or fail to capture inherent 3D structure underlying organization. To address these limitations, we present Graph Convolutional Model Epigenetic Regulation Gene Expression (GC-MERGE). Using graph-based framework, model incorporates...
We introduce a novel approach, STING, for spatial transcriptomic clustering analysis. Unlike existing state-of-the-art techniques that use graph-based neural networks (GNNs) trained on graphs generated from the proximity of tissue locations (or spots), STING incorporates spot-specific related genes. This feature allows to better distinguish between clusters and identify meaningful gene-gene relations knowledge discovery. It is nested GNN framework simultaneously models relations. Using gene...
Drug discovery is a complex and time-intensive process that requires identifying validating new therapeutic candidates. Computational approaches using large-scale biomedical knowledge graphs (KGs) offer promising solution to accelerate this process. However, extracting meaningful insights from KGs remains challenging due the complexity of graph traversal. Existing subgraph-based methods are tailored neural networks (GNNs), making them incompatible with other models, such as large language...