About

My research interests lie in machine learning, wireless communications and networking, and information theory. Current research topics include Transformers and large language models (LLMs), multi-armed bandits and reinforcement learning, privacy-preserving machine learning, federated learning and distributed/decentralized learning, machine learning for wireless communications and networking, and novel applications of artificial intelligence and machine learning.

Research topics

• Biology
• Genetics
• Cell biology
• Bioinformatics
• Cancer research
• Endocrinology
• Internal medicine
• Medicine
• Computational biology

Selected publications

• STCS: A Platform-Agnostic Framework for Cell-Level Reconstruction in Sequencing-Based Spatial Transcriptomics

  bioRxiv (Cold Spring Harbor Laboratory) · 2026-03-02

  articleOpen accessSenior author

  Abstract Sequencing-based spatial transcriptomics technologies, including Visium HD and Stereo-seq, now enable transcriptome-wide profiling at subcellular resolution. However, these platforms generate measurements over spatially barcoded units rather than biologically segmented cells, creating a fundamental bottleneck for cell-centric analysis and interpretation. Robust recon-struction of coherent single-cell transcriptomes from high-density spatial bins remains an unresolved computational challenge. Here we present STCS (Spatial Transcriptomics Cell Segmentation) , a platform-agnostic framework that reconstructs cell-level gene expression profiles by integrating nuclei segmentation with a joint transcriptomic–spatial distance model. STCS is governed by two interpretable parameters and incorporates a reference-free parameter selection strategy based on internal stability and spatial coherence metrics, enabling adaptable deployment across tissue types and technologies without requiring matched ground-truth annotations. We benchmark STCS on a Visium HD human lung cancer dataset with matched Xenium-derived cell segmentation, enabling direct cell-level validation, and on high-resolution Stereo-seq mouse brain data to assess cross-platform generalizability. Across multiple evaluation dimensions—including cell-type agreement, spatial organization, gene-expression fidelity, and compositional accuracy—STCS achieves consistent improvements over existing methods while preserving biologically coherent spatial structure. As sequencing-based spatial transcriptomics is rapidly adopted across biomedical research, STCS provides a broadly applicable and open-source solution for reconstructing cell-resolved transcriptomes, facilitating more reliable downstream analyses and cross-platform integration.

  PublisherDOI

• Abstract 6914: STCS: Spatial transcriptomics cell segmentation outperforms existing methods on multiple slides.

  Cancer Research · 2026-04-03

  articleSenior author

  Abstract Spatial transcriptomics (ST) has long been recognized as an advanced technique that provides insights on spatial information beyond what can be obtained from single-cell RNA sequencing. However, widely used sequencing-based ST approaches cannot provide cell level data because their results are aggregated into discrete bins rather than assigned to individual cells. With the advent of Visium HD and other subcellular-resolution platforms, accurate cell segmentation has become essential for extracting biologically meaningful, cell-level information.Here, we present STCS (Spatial Transcriptomics Cell Segmentation), a segmentation framework tailored for high-resolution ST data. We benchmarked STCS against several existing methods—including STHD, bin2cell, and Space Ranger—using a slide with both Visium HD and Xenium results. Evaluation using ground-truth Xenium cell boundary annotations demonstrated that STCS delivers the best performance, achieving 40% accuracy in cell-type prediction and showing the lowest spatial chaos score, a metric that quantifies how spatially continuous clusters are. We also applied STCS to another Visium HD slide from mouse intestinal regeneration model which contains tissue from different time points after radiation. Compared to default Visium HD binning, STCSsegmented cells show clear transcriptional differences by timepoints and identify several rare immune cell types. As a result, downstream analyses such as spatial cell-cell interaction inference and regional pattern characterization can be done in cell level which include more cell types and more immune related pathways like JAK-STAT pathway with STCS. In addition, STCS is versatile and can be applied to other sequencing-based ST methods like Stereo-seq, which offers nanometer-scale resolution. And it’s also an open-source tool with adjustable parameters for different tissue types.In summary, STCS is a robust and flexible cell segmentation tool that provides a one-stop solution for deriving biologically meaningful, cell-level information from high-resolution sequencing-based ST datasets. Citation Format: Xinyu Hu, Fengwei Zhan, Lixia C. Wu, Jose Gonzalez, Chuhanwen Sun, Rachel Ofer, Tyler Tran, Michael Verzi, Jiekun Yang. STCS: Spatial transcriptomics cell segmentation outperforms existing methods on multiple slides [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2026; Part 1 (Regular Abstracts); 2026 Apr 17-22; San Diego, CA. Philadelphia (PA): AACR; Cancer Res 2026;86(7 Suppl):Abstract nr 6914.

  PublisherDOI

• Impaired nitrogenous waste clearance promotes hepatocellular carcinoma

  Science Advances · 2026-01-09 · 2 citations

  articleOpen access

  In mammals, hepatic urea cycle enzymes (UCEs) convert ammonia, the toxic nitrogenous waste, into urea for excretion. In hepatocellular carcinoma (HCC), UCE expression is often heterogeneously repressed, but its role in tumorigenesis is unclear. We show that, as in patients, UCE expression is markedly reduced in multiple HCC mouse models, including those driven by oncogenic c-MET/β-catenin, leading to impaired ammonia clearance, altered amino acid metabolism, and increased pyrimidine synthesis. In contrast, UCE expression is largely preserved in c-MET/sgAxin1 tumors, allowing assessment of the consequences of UCE loss. Silencing individual UCEs increases ammonia burden and accelerates HCC with reprogrammed amino acid and pyrimidine metabolism, supporting a causal role for defective ammonia detoxification in oncogenesis. Notably, dietary protein restriction lowers hepatic ammonia and slows tumor growth. These findings establish a mechanistic link between nitrogen overload and hepatocarcinogenesis and highlight protein restriction as a feasible therapeutic strategy for patients with impaired nitrogenous waste handling.

  PublisherDOI

• Abstract 4914: Defining the immunoregulatory molecular drivers of dendritic cells using CRISPR screens in ex-vivo modelsvivo models

  Cancer Research · 2026-04-03

  articleSenior author

  Abstract Conventional dendritic cell type 1 (cDC1) plays a central and indispensable role in coordinating antitumor T-cell responses by producing cytokines such as IL-12 and expressing key regulatory molecules including PD-L1. Despite their importance in shaping antitumor immunity and influencing responses to modern immunotherapies, the molecular mechanisms controlling cDC1 activation and maturation remain poorly defined. This gap in knowledge is compounded by the extremely low abundance of cDC1 within the tumor microenvironment (TME), which limits the ability to experimentally interrogate their biology and apply genetic perturbation tools at scale. To address these challenges, we developed, optimized, and compared robust ex vivo differentiation systems capable of generating physiologically relevant cDC1-like cells from both murine and human hematopoietic stem cells (HSCs). Using a carefully designed two-stage culture platform incorporating Fms-like tyrosine kinase 3 ligand (Flt3L), DLL1-mediated Notch signaling, and granulocyte-macrophage colony-stimulating factor (GM-CSF), we efficiently produced CD103+ cDC1-like populations from mouse bone marrow progenitors and Clec9a+CD141+ cDC1-like cells from G-CSF-mobilized human peripheral blood CD34+ HSCs. These ex-vivo derived cells closely mirrored their in-vivo counterparts, exhibiting the expected surface immunophenotypes, high IL-12 secretion upon stimulation measured by ELISA, and strong functional responsiveness to TLR agonists and cell-associated antigens. Upon activation, cells consistently upregulated classical maturation markers including PD-L1, MHCII, and CD40, demonstrating their ability to undergo appropriate immune stimulation-driven maturation. To support downstream genetic studies, we optimized viral transduction strategies for primary HSCs, a major technical barrier in the field. An ecotropic retroviral system combined with retronectin substantially improved gene delivery efficiency in murine cells, while an amphotropic pseudotyped retroviral approach enhanced transduction of human CD34+ HSCs. These advances enable reliable introduction of CRISPR-based perturbations into precursor cells prior to differentiation. Together, this scalable, high-yield differentiation and gene-delivery platform provides a powerful foundation for conducting CRISPR-based genetic screens aimed at identifying the molecular regulators orchestrating cDC1 activation and maturation. These insights have the potential to inform the development of next-generation immunotherapies that harness or enhance dendritic cell-driven antitumor immunity. Defining the immunoregulatory molecular drivers of dendritic cells using CRISPR screens in ex-vivo models Citation Format: Mukta Asnani, Jiekun Yang. Defining the immunoregulatory molecular drivers of dendritic cells using CRISPR screens in ex-vivo modelsvivo models [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2026; Part 1 (Regular Abstracts); 2026 Apr 17-22; San Diego, CA. Philadelphia (PA): AACR; Cancer Res 2026;86(7 Suppl):Abstract nr 4914.

  PublisherDOI

• Epigenetically constrained astrocyte states underlie prefrontal cortex vulnerability in Down syndrome–associated Alzheimer’s disease

  bioRxiv (Cold Spring Harbor Laboratory) · 2026-04-21

  articleOpen accessSenior author

  Down syndrome (DS), caused by trisomy 21, confers a near-universal risk for Alzheimer's disease (AD), yet individuals exhibit marked variability in cognitive decline, suggesting the presence of cellular mechanisms that modulate vulnerability and resilience. However, these mechanisms remain poorly defined in the human brain. Here, we integrate matched single-nucleus RNA-seq and ATAC-seq profiles from the prefrontal cortex (PFC) and amygdala (AMY) of age-matched individuals with DS with and without AD (DSAD), enabling direct comparison within a shared genetic background. We identify basal astrocytes in the PFC as a selectively vulnerable cell state in DSAD, characterized by both reduced abundance and coordinated transcriptional and regulatory reprogramming. This state exhibits a shift away from homeostatic support functions, with decreased cytokine signaling and lipid-handling programs, alongside increased steroid- and nuclear receptor-associated activity. Concomitantly, chromatin accessibility profiling reveals reduced engagement of immune- and stress-responsive transcription factor programs, including AP-1, STAT, and BACH families, with linked regulatory perturbations at loci such as ABCA1, DAB2IP, and IL1RAP. Together, these findings define a previously unrecognized astrocyte state marked by epigenetic constraint and diminished responsiveness to stress and inflammatory signals, distinguishing it from classical reactive astrocyte phenotypes. Our results nominate PFC basal astrocytes as a key locus of vulnerability in DSAD and suggest that failure to mount appropriate astrocyte responses, rather than overt activation alone, may contribute to neurodegenerative progression.

  PublisherDOI

• Optimized Ex Vivo Differentiation of CD103 ⁺ Dendritic Cells and High-Efficiency Retroviral Transduction of Mouse Bone Marrow HSCs

  bioRxiv (Cold Spring Harbor Laboratory) · 2025-09-23 · 1 citations

  preprintOpen accessSenior authorCorresponding

  Abstract CD103 + conventional dendritic cells (cDC1s) are key drivers of antitumor immunity, but their scarcity and resistance to genetic manipulation make them difficult to study. We optimized a two-stage ex vivo culture system using key cytokines and growth factors to efficiently generate CD103 + cDC1-like cells from mouse bone marrow progenitors. These cells closely mimicked their in-vivo counterparts, displaying CD103 expression, robust cytokine production, and functional responses to immune stimulation. Additionally, we established a high-efficiency retroviral transduction method using ecotropic pseudotyped virus and retronectin-coated plates, significantly improving gene delivery into mouse hematopoietic stem cells. This integrated platform provides a powerful approach for dissecting CD103 + cDC1 biology and advancing dendritic cell–based immunotherapy research.

  PublisherOA PDFDOI

• A Spatial and Temporal Transcriptomic Atlas of Mouse Intestinal Regeneration

  bioRxiv (Cold Spring Harbor Laboratory) · 2025-08-06 · 1 citations

  preprintOpen access

  Background & Aims: The intestinal epithelium exhibits a remarkable capacity for regeneration following injury. However, the spatial and temporal dynamics of the injury-repair cycle remain incompletely understood. Methods: We employ spatial transcriptomics to create an atlas of the damage and repair response to ionizing radiation in the mouse intestine. We map molecular events driving epithelial recovery over a six-day period and 23 biological samples, spanning the early apoptotic response to tissue remodeling and repair. Results: The datasets capture mRNA of 19,042 genes in ∼26 million bins at 2µm resolution. Analysis revealed transcriptional patterns and niche signals that would remain undetected in bulk or single-cell approaches, including a non-random activation of interferon-target genes. Temporal shifts in cytokine and growth factor gene expression, particularly in the crypt and lower villus regions, corroborate published studies and reveal new predictions of the mechanisms governing intestinal healing. Global transcriptional upregulation was observed in the regenerating epithelium, suggesting hypertranscription is a hallmark of intestinal repair. Furthermore, we observe altered cellular differentiation trajectories and villus patterning at the early stages of regeneration. Conclusions: Together, our work provides a detailed spatiotemporal map of intestinal regeneration at subcellular resolution and nearly whole-genome scale. These data lay the groundwork for future discoveries and therapeutic strategies to enhance epithelial repair in inflammatory bowel diseases and other gastrointestinal pathologies or in response to side-effects of cancer therapies.

  PublisherOA PDFDOI

• Specific oncogene activation of the cell of origin in mucosal melanoma

  Nature Communications · 2025-07-22 · 1 citations

  articleOpen access

  Mucosal melanoma (MM) is a deadly cancer derived from mucosal melanocytes. To test the consequences of MM genetics, we develop a zebrafish model in which all melanocytes experience CCND1 expression and loss of PTEN and TP53. Surprisingly, melanoma only develops from melanocytes lining internal organs, analogous to the location of patient MM. We find that zebrafish MMs have a unique chromatin landscape from cutaneous melanomas. Internal melanocytes are labeled using a MM-specific transcriptional enhancer. Normal zebrafish internal melanocytes share a gene expression signature with MMs. Patient and zebrafish MMs show increased migratory neural crest and decreased antigen presentation gene expression, consistent with the increased metastatic behavior and decreased immunotherapy sensitivity of MM. Our work suggests that the cell state of the originating melanocyte influences the behavior of derived melanomas. Our animal model phenotypically and transcriptionally mimics patient tumors, allowing this model to be used for MM therapeutic discovery. As this is a non-MAPK driven genetically engineered model of melanoma, our work also has implications for the 15% of cutaneous melanoma patients who lack MAPK-driving mutations. Mucosal melanoma (MM) is a cancer with poor prognosis derived from mucosal melanocytes. Here, the authors implement a combination of genetic changes that occur in MM patients in a zebrafish model, revealing the potential MM cell of origin and showing that patient and zebrafish MMs share a gene signature that is more metastatic and immune-evasive.

  PublisherOA PDFDOI

• GRNFormer: A Biologically-Guided Framework for Integrating Gene Regulatory Networks into RNA Foundation Models

  ArXiv.org · 2025-03-03

  preprintOpen access

  Foundation models for single-cell RNA sequencing (scRNA-seq) have shown promising capabilities in capturing gene expression patterns. However, current approaches face critical limitations: they ignore biological prior knowledge encoded in gene regulatory relationships and fail to leverage multi-omics signals that could provide complementary regulatory insights. In this paper, we propose GRNFormer, a new framework that systematically integrates multi-scale Gene Regulatory Networks (GRNs) inferred from multi-omics data into RNA foundation model training. Our framework introduces two key innovations. First, we introduce a pipeline for constructing hierarchical GRNs that capture regulatory relationships at both cell-type-specific and cell-specific resolutions. Second, we design a structure-aware integration framework that addresses the information asymmetry in GRNs through two technical advances: (1) A graph topological adapter using multi-head cross-attention to weight regulatory relationships dynamically, and (2) a novel edge perturbation strategy that perturb GRNs with biologically-informed co-expression links to augment graph neural network training. Comprehensive experiments have been conducted on three representative downstream tasks across multiple model architectures to demonstrate the effectiveness of GRNFormer. It achieves consistent improvements over state-of-the-art (SoTA) baselines: $3.6\%$ increase in drug response prediction correlation, $9.6\%$ improvement in single-cell drug classification AUC, and $1.1\%$ average gain in gene perturbation prediction accuracy.

  PublisherOA PDFDOI

• Type 2 Diabetes and Obesity Alter Exercise Training-Induced Transcriptional Adaptations to Subcutaneous White Adipose Tissue

  bioRxiv (Cold Spring Harbor Laboratory) · 2025-10-12

  preprintOpen access

  Abstract White adipose tissue (WAT) dysfunction contributes to obesity-associated metabolic disease and type 2 diabetes (T2D). Rodent studies have demonstrated that exercise training improves WAT function, but molecular studies investigating exercise training effects on WAT in humans have been limited, particularly in the context of metabolic disease. Here, we defined the subcutaneous WAT (scWAT) transcriptome in middle-aged adults (10 male, 19 female) that were classified by lower BMI (<27 kg/m 2 ), higher BMI (≥27 kg/m 2 ), and T2D status before and after a 10-week endurance exercise regimen. At baseline, 624 genes were significantly upregulated and 112 genes downregulated in the scWAT from higher BMI participants compared to lower BMI. There was a spectrum of pathway dysregulation in scWAT in higher BMI individuals, ranging from increased markers of extracellular matrix (ECM) deposition and inflammation to decreased circadian rhythm gene expression. In people with T2D, there were additional transcriptomic differences such as translation-related pathways, selenoamino acid metabolism, and proteoglycans. Exercise training had robust effects on the transcriptome regardless of metabolic status, and notably, for the high BMI and T2D groups, training reversed several of the detrimental gene expression patterns in a cell-type-specific manner. These beneficial exercise-induced transcriptomic adaptations significantly correlated with lower levels of free fatty acids and blood pressure, particularly in participants with higher BMI and T2D. By integrating our exercise training-modulated genes with GWAS meta-analysis of physical activity, genes influenced by exercise training in the higher BMI group showed a significant enrichment in genetic associations of exercise traits in the population. A circadian rhythm-related transcription factor NR1D1 was enriched in enhancers linked with both the exercise differentially expressed genes (DEGs) and GWAS signals, suggesting a link between the circadian rhythm and training-induced adaptations. These findings demonstrate that obesity and T2D result in marked, progressive alterations in cell-type specific gene transcription in scWAT, while endurance exercise training reverses many of the metabolic disease-associated adaptations. Identification of novel molecular pathways regulated by exercise training can lead to therapeutic targets for obesity and metabolic disease.

  PublisherOA PDFDOI

Frequent coauthors

• Manolis Kellis

  Massachusetts Institute of Technology

  34 shared

• Mazhar Adli

  Indo-American Center

  26 shared

• Kelly P. Burke

  24 shared

• Amy Y. Huang

  23 shared

• David Liu

  Dana-Farber Cancer Institute

  23 shared

• Genevieve M. Boland

  22 shared

• Arlene H. Sharpe

  Massachusetts General Hospital

  19 shared

• Maria Vamvini

  Harvard University

  17 shared

Labs

• Jing Yang's LabPI

  Not provided

Education

• B.S., Electrical Engineering

  University of Science and Technology of China (USTC)

• M.S., Electrical Engineering

  University of Maryland, College Park

• Ph.D., Electrical Engineering

  University of Maryland, College Park

Awards & honors

• IEEE Fellow, Class of 2026
• Bose Dissertation Excellence Award