A5059011321- Genomics and Chromatin Dynamics
- Cancer Genomics and Diagnostics
- Radiomics and Machine Learning in Medical Imaging
- AI in cancer detection
- Single-cell and spatial transcriptomics
- Advanced Image and Video Retrieval Techniques
- Domain Adaptation and Few-Shot Learning
- RNA modifications and cancer
- Bioinformatics and Genomic Networks
- Genetics, Bioinformatics, and Biomedical Research
- Osteoarthritis Treatment and Mechanisms
- Total Knee Arthroplasty Outcomes
- Multimodal Machine Learning Applications
- RNA and protein synthesis mechanisms
- Machine Learning and Algorithms
- Cancer-related Molecular Pathways
- Cell Image Analysis Techniques
- Epigenetics and DNA Methylation
- Electrostatic Discharge in Electronics
- Value Engineering and Management
- Advancements in Semiconductor Devices and Circuit Design
- Angiogenesis and VEGF in Cancer
- Viral Infections and Vectors
- Mathematical Biology Tumor Growth
- Silicon Carbide Semiconductor Technologies
New York University
2019-2025
Institute for Systems Biology
2023-2025
Broad Institute
2025
Office of Science
2024
Courant Institute of Mathematical Sciences
2020
Nanjing Agricultural University
2020
Grinnell College
2015
Background The methods for assessing knee osteoarthritis (OA) do not provide enough comprehensive information to make robust and accurate outcome predictions. Purpose To develop a deep learning (DL) prediction model risk of OA progression by using radiographs in patients who underwent total replacement (TKR) matched control did undergo TKR. Materials Methods In this retrospective analysis that used data from the Initiative, DL on was developed predict both likelihood patient undergoing TKR...
Investigating how chromatin organization determines cell-type-specific gene expression remains challenging. Experimental methods for measuring three-dimensional organization, such as Hi-C, are costly and have technical limitations, restricting their broad application particularly in high-throughput genetic perturbations. We present C.Origami, a multimodal deep neural network that performs de novo prediction of using DNA sequence two genomic features-CTCF binding accessibility. C.Origami...
Transcriptional regulation, which involves a complex interplay between regulatory sequences and proteins, directs all biological processes. Computational models of transcription lack generalizability to accurately extrapolate unseen cell types conditions. Here we introduce GET (general expression transformer), an interpretable foundation model designed uncover grammars across 213 human fetal adult types1,2. Relying exclusively on chromatin accessibility data sequence information, achieves...
We characterized a prospective endometrial carcinoma (EC) cohort containing 138 tumors and 20 enriched normal tissues using 10 different omics platforms. Targeted quantitation of two peptides can predict antigen processing presentation machinery activity, may inform patient selection for immunotherapy. Association analysis between MYC activity metformin treatment in both patients cell lines suggests potential role non-diabetic with elevated activity. PIK3R1 in-frame indels are associated AKT...
Knee osteoarthritis (OA) is a chronic degenerative disorder of joints and it the most common reason leading to total knee joint replacement. Diagnosis OA involves subjective judgment on symptoms, medical history, radiographic readings using Kellgren-Lawrence grade (KL-grade). Deep learning-based methods such as Convolution Neural Networks (CNN) have recently been applied automatically diagnose OA. In this study, we Residual Network (ResNet) first detect from radiographs later combine ResNet...
We introduce a pioneering approach that integrates pathology imaging with transcriptomics and proteomics to identify predictive histology features associated critical clinical outcomes in cancer. utilize 2,755 H&E-stained histopathological slides from 657 patients across 6 cancer types CPTAC. Our models effectively recapitulate distinctions readily made by human pathologists: tumor vs. normal (AUROC = 0.995) tissue-of-origin 0.979). further investigate power on tasks not normally performed...
The randomized or cross-validated split of training and testing sets has been adopted as the gold standard machine learning for decades. establishment these protocols are based on two assumptions: (i)-fixing dataset to be eternally static so we could evaluate different algorithms models; (ii)-there is a complete set annotated data available researchers industrial practitioners. However, in this article, intend take closer critical look at protocol itself point out its weakness limitation,...
Brain tumors affect about 1 million people in the U.S., with aggressive types like glioblastoma having very low survival rates due to complex tumor biology and protective blood-brain barrier. Current treatments are limited effectiveness, our understanding of brain remains incomplete. High dimensional multiplexed imaging has enabled us better understand microenvironment (TME); however, analyses typically rely on cell segmentation, which is error-prone, may discard useful context outside...
Abstract Despite advancements in deep learning for histopathology, integrating these insights with multi-omics data to uncover clinically relevant omics pathway-level signatures remains a challenge. Our study addresses this gap by applying unsupervised techniques on pan-cancer data, leveraging 3,080 Hematoxylin and Eosin (H&E) images from 1,010 patients Clinical Proteomic Tumor Analysis Consortium (CPTAC) that drive discernable morphology phenotypes at the tissue level. First, imaging...
Abstract The mammalian genome is spatially organized in the nucleus to enable cell type-specific gene expression. Investigating how chromatin organization determines this specificity remains a challenge. Methods for measuring 3D organization, such as Hi-C, are costly and bear strong technical limitations, restricting their broad application particularly high-throughput genetic perturbations. In study, we present C.Origami, deep neural network model that performs de novo prediction of...
Transcriptional regulation, involving the complex interplay between regulatory sequences and proteins, directs all biological processes. Computational models of transcription lack generalizability to accurately extrapolate in unseen cell types conditions. Here, we introduce GET, an interpretable foundation model designed uncover grammars across 213 human fetal adult types. Relying exclusively on chromatin accessibility data sequence information, GET achieves experimental-level accuracy...
The interaction between tumors and their microenvironment is complex heterogeneous. Recent developments in high-dimensional multiplexed imaging have revealed the spatial organization of tumor tissues at molecular level. However, discovery thorough characterization (TME) remains challenging due to scale complexity images. Here, we propose a self-supervised representation learning framework, CANVAS, that enables novel types TMEs. CANVAS vision transformer directly takes images trained using...
Rendering Wikipedia content through mobile and augmented reality mediums can enable new forms of interaction in urban-focused user communities facilitating learning, communication knowledge exchange. With this objective mind, work we develop a application that allows for the recognition notable sites featured on Wikipedia. The is powered by deep neural network has been trained crowd-sourced imagery describing interest, such as buildings, statues, museums or other physical entities are...
ABSTRACT PCV2 and SS2 were clinically two major pathogens in pigs they zoonotic pathogens. There was extensive cellular tropism for both SS2, as well as, PK15 cells infection mainly cell model. It found that when infected before at the MOI=0.1, could cause more damage to cells. ITRAQ labeling proteomic technology used explore differentially expressed proteome of single co-infection The results showed there total 4736 proteins distinct changed this models aggravated directly big amount like...
Osteoarthritis (OA) is a chronic degenerative disorder of joints and the most common reason leading to total knee joint replacement (TKR). In this paper, we implemented semi-supervised learning approach based on Unsupervised Data Augmentation (UDA) along with valid perturbations for radiographs enhance performance supervised TKR outcome prediction model. Our results suggest that use provides superior compared (AUC 0.79 ± 0.04 vs 0.74 0.04).
Being able to automatically recognize notable sites in the physical world using artificial intelligence embedded mobile devices can pave way new forms of urban exploration and open novel channels interactivity between residents, travellers cities. Although development outdoor recognition systems has been a topic interest for while, most works have limited geographic coverage due lack high quality image data that be used training site engines. As result, prior usually generality operate on...