Jiarui Ding
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Previous Organizations / Institutes
The organizations / institutes from the scholar's other articles in the past ten years
Jiarui Ding has a total of 98 co-authors. Here are the top co-authors by number of shared publications:
- Aviv Regev
Department of Biology, Massachusetts Institute of Technology, Cambridge, USA
6 shared publications Carlos Vasquez-Rios
Department of Computer Science, University of British Columbia, Vancouver, BC V6T 1Z4, Canada
4 shared publicationsAnne Condon
Department of Computer Science, University of British Columbia, Vancouver, BC V6T 1Z4, Canada
2 shared publications- Anne Condon
University of British Columbia
2 shared publications Aviv Regev
Broad Institute of MIT and Harvard, Cambridge, MA, USA
2 shared publicationsChen Wang
F.M. Kirby Neurobiology Center, Boston Children’s Hospital, Harvard Medical School, Boston, USA
2 shared publications
Latest Publications
- Open Access
Computational identification of antigen-specific T cell groups through generative epitope modeling
- Top Cited PapersOpen Access
- 1 May 2026
- Published by Springer Nature in Nature Biotechnology
- p. 1-8
TxPert: using multiple knowledge graphs for prediction of transcriptomic perturbation effects
Accurately predicting cellular responses to genetic perturbations is essential for understanding disease mechanisms and designing effective therapies. Yet, exhaustively exploring the space of possible perturbations (for example, multigene perturbations or across tissues and cell types) is prohibitively expensive, motivating methods that can generalize to unseen conditions. We present TxPert, a latent-transfer-based deep learning method that uses multiple knowledge graphs of gene (product)–gene (product) relationships to predict transcriptomic perturbation effects. Different knowledge graphs encode complementary information and we show that a combination of graphs derived from biological databases and high-throughput perturbation screens yields the best performance. For predictions of single unseen perturbations, TxPert approaches the performance of split-half experimental reproducibility. For double unseen perturbations and single perturbations in a different cell line, its predictions increase Person Δ for unseen single perturbations by 8–25% over existing methods.
- 1 May 2026
- Published by Springer Nature in Nature Biotechnology
- p. 1-8
Top Cited PapersOpen AccessTxPert: using multiple knowledge graphs for prediction of transcriptomic perturbation effects
Accurately predicting cellular responses to genetic perturbations is essential for understanding disease mechanisms and designing effective therapies. Yet, exhaustively exploring the space of possible perturbations (for example, multigene perturbations or across tissues and cell types) is prohibitively expensive, motivating methods that can generalize to unseen conditions. We present TxPert, a latent-transfer-based deep learning method that uses multiple knowledge graphs of gene (product)–gene (product) relationships to predict transcriptomic perturbation effects. Different knowledge graphs encode complementary information and we show that a combination of graphs derived from biological databases and high-throughput perturbation screens yields the best performance. For predictions of single unseen perturbations, TxPert approaches the performance of split-half experimental reproducibility. For double unseen perturbations and single perturbations in a different cell line, its predictions increase Person Δ for unseen single perturbations by 8–25% over existing methods.
- Open Access
- 16 April 2026
- Published by Elsevier in Cell Genomics
- Vol. 6 (6) , 101217
ProtoCloud: A prototypical self-explaining model for single-cell analysis
- 16 April 2026
- Published by Elsevier in Cell Genomics
- Vol. 6 (6) , 101217
Open AccessProtoCloud: A prototypical self-explaining model for single-cell analysis
- Open Access
ProtoCloud: a Prototypical Self-explaining Model for Single-cell Analysis
Cell type annotation is a fundamental task in single-cell genomics. Although various methods have been developed for automatic cell type annotation, they often function as black-box models, making predictions without explaining their reasoning and lacking proper uncertainty estimation for their predictions. Furthermore, they often struggle to annotate rare cell types. We introduce ProtoCloud, a self-explaining deep generative model trained end-to-end to embed cells into a structured, low-dimensional space organized around cell type-specific prototypes. Coupled with a specifically-designed data augmentation strategy, it matches or outperforms existing methods in cell type annotation across 11 large-scale datasets, particularly for rare cell types. Moreover, ProtoCloud improves data annotation quality by identifying and re-annotating mis-annotated training cells through a built-in certainty quantification mechanism based on cell-prototype similarity. Finally, ProtoCloud provides interpretable predictions by identifying key genes that drive its classifications, facilitating the discovery of both known and novel cell type marker genes. Applied to a time-course dataset of post-injury retinal neurons, ProtoCloud successfully annotates previously unassigned cells; on the esophageal cell atlas, it identifies rare but potentially important cell populations and their marker genes relevant to esophageal inflammation.
Open AccessProtoCloud: a Prototypical Self-explaining Model for Single-cell Analysis
Cell type annotation is a fundamental task in single-cell genomics. Although various methods have been developed for automatic cell type annotation, they often function as black-box models, making predictions without explaining their reasoning and lacking proper uncertainty estimation for their predictions. Furthermore, they often struggle to annotate rare cell types. We introduce ProtoCloud, a self-explaining deep generative model trained end-to-end to embed cells into a structured, low-dimensional space organized around cell type-specific prototypes. Coupled with a specifically-designed data augmentation strategy, it matches or outperforms existing methods in cell type annotation across 11 large-scale datasets, particularly for rare cell types. Moreover, ProtoCloud improves data annotation quality by identifying and re-annotating mis-annotated training cells through a built-in certainty quantification mechanism based on cell-prototype similarity. Finally, ProtoCloud provides interpretable predictions by identifying key genes that drive its classifications, facilitating the discovery of both known and novel cell type marker genes. Applied to a time-course dataset of post-injury retinal neurons, ProtoCloud successfully annotates previously unassigned cells; on the esophageal cell atlas, it identifies rare but potentially important cell populations and their marker genes relevant to esophageal inflammation.
- Open Access
- 1 December 2025
- Published by Elsevier in Cell Genomics
- Vol. 6 (2) , 101073
CellUntangler: Separating distinct biological signals in single-cell data with deep generative models
- 1 December 2025
- Published by Elsevier in Cell Genomics
- Vol. 6 (2) , 101073
Open AccessCellUntangler: Separating distinct biological signals in single-cell data with deep generative models
- Open Access
- 28 October 2025
- Published by Elsevier in Cell Reports Methods
- Vol. 5 (11) , 101211
Single-cell multiomics data integration and generation with scPairing
- 28 October 2025
- Published by Elsevier in Cell Reports Methods
- Vol. 5 (11) , 101211
Open AccessSingle-cell multiomics data integration and generation with scPairing
- Open Access
Repertoire-level generation of T-cell epitopes with a large-scale generative transformer
Single-cell TCR sequencing enables high-resolution analysis of T Cell Receptor (TCR) diversity and clonality, offering valuable insights into immune responses and disease mechanisms. However, identifying cognate epitopes for individual TCRs requires complex and costly functional assays. We address this challenge with EpitopeGen, a large-scale transformer model based on the GPT-2 architecture that generates potential cognate epitope sequences directly from TCR sequences. To overcome the scarcity of TCR-epitope binding pairs (≈ 100,000), EpitopeGen uses a semi-supervised learning method, termed BINDSEARCH, which searches over 70 billion potential pairs and incorporates high binding affinity predictions as pseudo-labels. To incorporate CD8 + T cell biology into the model as an inductive bias, EpitopeGen employs a novel data balancing method, termed Antigen Category Filter, that carefully controls antigen category ratios in its training dataset. EpitopeGen significantly outperforms baseline approaches, generating epitopes with high binding affinity, diversity, naturalness, and biophysical stability. Notably, the epitopes generated by EpitopeGen follow biologically plausible antigen category distributions, a crucial feature not achieved by other models. Using EpitopeGen, we directly identify subsets of clonally expanded tumor-infiltrating lymphocytes that recognize tumor-associated antigens, exhibiting elevated cytotoxicity and reduced exhaustion markers. From COVID-19 patients, EpitopeGen detects T cells that recognize COVID-19 spike proteins and non-structural proteins with distinct transcriptomic characteristics. In conclusion, EpitopeGen represents the first computational method that enables direct inference of antigen recognition profiles of CD8 + T cells from plain TCR repertoires.
Open AccessRepertoire-level generation of T-cell epitopes with a large-scale generative transformer
Single-cell TCR sequencing enables high-resolution analysis of T Cell Receptor (TCR) diversity and clonality, offering valuable insights into immune responses and disease mechanisms. However, identifying cognate epitopes for individual TCRs requires complex and costly functional assays. We address this challenge with EpitopeGen, a large-scale transformer model based on the GPT-2 architecture that generates potential cognate epitope sequences directly from TCR sequences. To overcome the scarcity of TCR-epitope binding pairs (≈ 100,000), EpitopeGen uses a semi-supervised learning method, termed BINDSEARCH, which searches over 70 billion potential pairs and incorporates high binding affinity predictions as pseudo-labels. To incorporate CD8 + T cell biology into the model as an inductive bias, EpitopeGen employs a novel data balancing method, termed Antigen Category Filter, that carefully controls antigen category ratios in its training dataset. EpitopeGen significantly outperforms baseline approaches, generating epitopes with high binding affinity, diversity, naturalness, and biophysical stability. Notably, the epitopes generated by EpitopeGen follow biologically plausible antigen category distributions, a crucial feature not achieved by other models. Using EpitopeGen, we directly identify subsets of clonally expanded tumor-infiltrating lymphocytes that recognize tumor-associated antigens, exhibiting elevated cytotoxicity and reduced exhaustion markers. From COVID-19 patients, EpitopeGen detects T cells that recognize COVID-19 spike proteins and non-structural proteins with distinct transcriptomic characteristics. In conclusion, EpitopeGen represents the first computational method that enables direct inference of antigen recognition profiles of CD8 + T cells from plain TCR repertoires.
- Open Access
CellUntangler: separating distinct biological signals in single-cell data with deep generative models
Single-cell RNA-seq data have provided new insights into intracellular and intercellular processes. Because multiple processes are active in each cell simultaneously, such as its cell type program, differentiation, the cell cycle, and environmental responses, their respective signals can confound one another, requiring methods that can separate and filter different complex biological signals. Each such signal is based on different gene activities and can define different relationships between cells. However, existing methods often focus on a single process or rely on overly restrictive assumptions, thus removing, rather than disentangling biological signals. Here, we develop CellUntangler, a deep generative model that embeds cells into a flexible latent space composed of multiple subspaces, each designed with an appropriate geometry to capture a distinct signal. We apply CellUntangler to datasets containing only cycling cells and both cycling and non-cycling cells, generating embeddings in which the cell cycle signal is disentangled from non-cell cycle specific signals, such as cell type or differentiation trajectory. We demonstrate CellUntangler’s extensibility by using it to capture and separate spatial from non-spatial signals. With CellUntangler, we can obtain latent embeddings that capture various biological signals and perform enhancement or filtering at the gene expression level for downstream analyses.
Open AccessCellUntangler: separating distinct biological signals in single-cell data with deep generative models
Single-cell RNA-seq data have provided new insights into intracellular and intercellular processes. Because multiple processes are active in each cell simultaneously, such as its cell type program, differentiation, the cell cycle, and environmental responses, their respective signals can confound one another, requiring methods that can separate and filter different complex biological signals. Each such signal is based on different gene activities and can define different relationships between cells. However, existing methods often focus on a single process or rely on overly restrictive assumptions, thus removing, rather than disentangling biological signals. Here, we develop CellUntangler, a deep generative model that embeds cells into a flexible latent space composed of multiple subspaces, each designed with an appropriate geometry to capture a distinct signal. We apply CellUntangler to datasets containing only cycling cells and both cycling and non-cycling cells, generating embeddings in which the cell cycle signal is disentangled from non-cell cycle specific signals, such as cell type or differentiation trajectory. We demonstrate CellUntangler’s extensibility by using it to capture and separate spatial from non-spatial signals. With CellUntangler, we can obtain latent embeddings that capture various biological signals and perform enhancement or filtering at the gene expression level for downstream analyses.
- Open Access
Single-cell multiomics data integration and generation with scPairing
Single-cell multiomics technologies generate paired measurements of different cellular modalities, such as gene expression and chromatin accessibility. However, multiomics technologies are more expensive than their unimodal counterparts, resulting in smaller and fewer available multiomics datasets. Here, we present scPairing, a deep learning model inspired by Contrastive Language-Image Pre-Training, which embeds different modalities from the same single cells onto a common embedding space. We leverage the common embedding space to generate novel multiomics data following bridge integration, a method that uses an existing multiomics bridge to link unimodal data. Through extensive benchmarking, we show that scPairing constructs an embedding space that fully captures both coarse and fine biological structures. We then use scPairing to generate new multiomics data from retina, immune, and renal cells. Furthermore, we extend scPairing to generate trimodal data. Researchers can use these generated multiomics datasets to discover new biological relationships across modalities or confirm existing hypotheses.
Open AccessSingle-cell multiomics data integration and generation with scPairing
Single-cell multiomics technologies generate paired measurements of different cellular modalities, such as gene expression and chromatin accessibility. However, multiomics technologies are more expensive than their unimodal counterparts, resulting in smaller and fewer available multiomics datasets. Here, we present scPairing, a deep learning model inspired by Contrastive Language-Image Pre-Training, which embeds different modalities from the same single cells onto a common embedding space. We leverage the common embedding space to generate novel multiomics data following bridge integration, a method that uses an existing multiomics bridge to link unimodal data. Through extensive benchmarking, we show that scPairing constructs an embedding space that fully captures both coarse and fine biological structures. We then use scPairing to generate new multiomics data from retina, immune, and renal cells. Furthermore, we extend scPairing to generate trimodal data. Researchers can use these generated multiomics datasets to discover new biological relationships across modalities or confirm existing hypotheses.
- Top Cited PapersOpen Access
- 18 April 2024
- Published by Springer Nature in Nature Communications
- Vol. 15 (1) , 1-18
An esophagus cell atlas reveals dynamic rewiring during active eosinophilic esophagitis and remission
Coordinated cell interactions within the esophagus maintain homeostasis, and disruption can lead to eosinophilic esophagitis (EoE), a chronic inflammatory disease with poorly understood pathogenesis. We profile 421,312 individual cells from the esophageal mucosa of 7 healthy and 15 EoE participants, revealing 60 cell subsets and functional alterations in cell states, compositions, and interactions that highlight previously unclear features of EoE. Active disease displays enrichment of ALOX15+ macrophages, PRDM16+ dendritic cells expressing the EoE risk gene ATP10A, and cycling mast cells, with concomitant reduction of TH17 cells. Ligand–receptor expression uncovers eosinophil recruitment programs, increased fibroblast interactions in disease, and IL-9+IL-4+IL-13+ TH2 and endothelial cells as potential mast cell interactors. Resolution of inflammation-associated signatures includes mast and CD4+ TRM cell contraction and cell type-specific downregulation of eosinophil chemoattractant, growth, and survival factors. These cellular alterations in EoE and remission advance our understanding of eosinophilic inflammation and opportunities for therapeutic intervention.
- 18 April 2024
- Published by Springer Nature in Nature Communications
- Vol. 15 (1) , 1-18
Top Cited PapersOpen AccessAn esophagus cell atlas reveals dynamic rewiring during active eosinophilic esophagitis and remission
Coordinated cell interactions within the esophagus maintain homeostasis, and disruption can lead to eosinophilic esophagitis (EoE), a chronic inflammatory disease with poorly understood pathogenesis. We profile 421,312 individual cells from the esophageal mucosa of 7 healthy and 15 EoE participants, revealing 60 cell subsets and functional alterations in cell states, compositions, and interactions that highlight previously unclear features of EoE. Active disease displays enrichment of ALOX15+ macrophages, PRDM16+ dendritic cells expressing the EoE risk gene ATP10A, and cycling mast cells, with concomitant reduction of TH17 cells. Ligand–receptor expression uncovers eosinophil recruitment programs, increased fibroblast interactions in disease, and IL-9+IL-4+IL-13+ TH2 and endothelial cells as potential mast cell interactors. Resolution of inflammation-associated signatures includes mast and CD4+ TRM cell contraction and cell type-specific downregulation of eosinophil chemoattractant, growth, and survival factors. These cellular alterations in EoE and remission advance our understanding of eosinophilic inflammation and opportunities for therapeutic intervention.