Han Wen
Han Wen has a total of 297 co-authors. Here are the top co-authors by number of shared publications:
Qinrui Wang
DP Technology, Beijing, China
7 shared publications- Wenjun Zheng
University at Buffalo
6 shared publications Ruichu Gu
AI for Science Institute, Beijing, China
5 shared publications- Fan Yang
Kidney Disease Center of the First Affiliated Hospital and Department of Biophysics, Zhejiang University School of Medicine, Hangzhou, China
4 shared publications - Gabriela K. Popescu
Department of Biochemistry, Jacobs School of Medicine and Biomedical Sciences, University at Buffalo, Buffalo, USA
4 shared publications - Gary J. Iacobucci
University at Buffalo
4 shared publications
Latest Publications
- 9 September 2026
- Published by Springer Nature in Nature
- Vol. 658 (8137) , 1066-1076
An operational perturbation proteomics-based virtual cell model
Artificial intelligence-empowered virtual cell models represent an emerging approach for in silico drug discovery1,2,3, yet most existing approaches lack large-scale, time-resolved perturbation proteomics data and interpretable frameworks for predicting therapeutic responses. Here we generated more than 38 million temporal protein-abundance measurements from systematically perturbed breast cancer cell lines, and developed ProteinTalks, a virtual cell model. Central to ProteinTalks is the synergy of this large-scale dynamic proteomic resource and the model architecture, enabling a new pretraining framework that learns transferable dynamical latent representations from temporal proteome trajectories. By modelling how proteins respond conditionally to different perturbations, this approach enables the model to function as an operational tool for diverse drug discovery tasks: predicting drug efficacy and synergy, discovering new drug combinations, probing proteins associated with drug resistance, stratifying patient responses and prioritizing drug candidates for patient organoids. It also shows robust transferability, extending beyond cell lines to patient-derived organoids and clinical biopsies, generally achieving higher performance than the selected benchmark implementations under the evaluated protocols. Together, ProteinTalks shows how scalable pretraining of transferable dynamic representations enables operational, dynamics-aware, proteomics-based virtual cell models to advance in silico drug discovery.
- 9 September 2026
- Published by Springer Nature in Nature
- Vol. 658 (8137) , 1066-1076
An operational perturbation proteomics-based virtual cell model
Artificial intelligence-empowered virtual cell models represent an emerging approach for in silico drug discovery1,2,3, yet most existing approaches lack large-scale, time-resolved perturbation proteomics data and interpretable frameworks for predicting therapeutic responses. Here we generated more than 38 million temporal protein-abundance measurements from systematically perturbed breast cancer cell lines, and developed ProteinTalks, a virtual cell model. Central to ProteinTalks is the synergy of this large-scale dynamic proteomic resource and the model architecture, enabling a new pretraining framework that learns transferable dynamical latent representations from temporal proteome trajectories. By modelling how proteins respond conditionally to different perturbations, this approach enables the model to function as an operational tool for diverse drug discovery tasks: predicting drug efficacy and synergy, discovering new drug combinations, probing proteins associated with drug resistance, stratifying patient responses and prioritizing drug candidates for patient organoids. It also shows robust transferability, extending beyond cell lines to patient-derived organoids and clinical biopsies, generally achieving higher performance than the selected benchmark implementations under the evaluated protocols. Together, ProteinTalks shows how scalable pretraining of transferable dynamic representations enables operational, dynamics-aware, proteomics-based virtual cell models to advance in silico drug discovery.
- Open Access
- 27 July 2026
- Published by Springer Nature in Genome Biology
- Vol. 27 (1) , 236
IDEAL-Age: an interpretable deep learning framework for single-cell resolution profiling of immunological aging
Immunosenescence increases susceptibility to infection and reduces vaccine responsiveness, yet bulk transcriptomic clocks obscure the cellular heterogeneit
- 27 July 2026
- Published by Springer Nature in Genome Biology
- Vol. 27 (1) , 236
Open AccessIDEAL-Age: an interpretable deep learning framework for single-cell resolution profiling of immunological aging
Immunosenescence increases susceptibility to infection and reduces vaccine responsiveness, yet bulk transcriptomic clocks obscure the cellular heterogeneit
- 22 July 2026
- Published by Springer Nature in Nature
- Vol. 656 (8127) , 463-473
Precise DNA base editing using AlphaFold3-based contact modelling
- 22 July 2026
- Published by Springer Nature in Nature
- Vol. 656 (8127) , 463-473
Precise DNA base editing using AlphaFold3-based contact modelling
- Open Access
- 21 July 2026
- Published by Springer Nature in Research Square
Generating and navigating single cell dynamics via a geodesic bridge between nonlinear transcriptional and linear latent manifolds
Time-series single-cell RNA sequencing (scRNA-seq) captures cellular processes as sparse and unpaired snapshots, limiting our ability not only to reconstruct continuous cell state transitions, but also to navigate between states in a controlled and interpretable manner. Here we present GeoBridge, a framework that formulates biological-time interpolation as a Riemannian metric-learning problem.. GeoBridge learns an invertible geodesic-bridge mapping, defines a simple Euclidean geometry in the latent space, and induces a Riemannian metric on the transcriptional manifold through the corresponding pullback metric. Under this learned metric, cellular dynamics are represented as geodesic trajectories affinely parameterized by biological time. By constraining temporal transitions to follow constant-velocity straight-line geodesics in the latent Euclidean space, GeoBridge enables explicit and closed-form interpolation of unobserved intermediate cell states, enabling reconstruction of unobserved intermediate states and efficient navigation between distinct cellular phenotypes at a single-cell resolution. By mapping interpolated trajectories back to the original gene expression space, GeoBridge recovers smooth transcriptional programs that are robust to noise and snapshot sparsity. Leveraging the derived geodesic potentials, GeoBridge further infers pseudo-temporal trajectories with superior fidelity compared to mainstream approaches from single-snapshot scRNA-seq data without temporal annotation, and directly identifies genes that drive progression along transition paths. We demonstrate the utility of GeoBridge across diverse systems, where GeoBridge resolves EMT–MET progression in cancer stem cells and identifies stage-specific modules as well as the branching cell-fate dynamics in human pluripotent stem cells with higher reconstruction accuracy than state-of-the-art methods. More importantly, GeoBridge supports single-cell fate navigation in multi-target hematopoietic lineages, allowing neutrophil-biased cells to be virtually guided toward mast-cell fates along biologically plausible paths. Together, GeoBridge establishes a principled pullback-geometric method that transforms sparse single-cell measurements into a continuous, controllable landscape for the reconstruction, navigation and manipulation of cellular state transitions.
- 21 July 2026
- Published by Springer Nature in Research Square
Open AccessGenerating and navigating single cell dynamics via a geodesic bridge between nonlinear transcriptional and linear latent manifolds
Time-series single-cell RNA sequencing (scRNA-seq) captures cellular processes as sparse and unpaired snapshots, limiting our ability not only to reconstruct continuous cell state transitions, but also to navigate between states in a controlled and interpretable manner. Here we present GeoBridge, a framework that formulates biological-time interpolation as a Riemannian metric-learning problem.. GeoBridge learns an invertible geodesic-bridge mapping, defines a simple Euclidean geometry in the latent space, and induces a Riemannian metric on the transcriptional manifold through the corresponding pullback metric. Under this learned metric, cellular dynamics are represented as geodesic trajectories affinely parameterized by biological time. By constraining temporal transitions to follow constant-velocity straight-line geodesics in the latent Euclidean space, GeoBridge enables explicit and closed-form interpolation of unobserved intermediate cell states, enabling reconstruction of unobserved intermediate states and efficient navigation between distinct cellular phenotypes at a single-cell resolution. By mapping interpolated trajectories back to the original gene expression space, GeoBridge recovers smooth transcriptional programs that are robust to noise and snapshot sparsity. Leveraging the derived geodesic potentials, GeoBridge further infers pseudo-temporal trajectories with superior fidelity compared to mainstream approaches from single-snapshot scRNA-seq data without temporal annotation, and directly identifies genes that drive progression along transition paths. We demonstrate the utility of GeoBridge across diverse systems, where GeoBridge resolves EMT–MET progression in cancer stem cells and identifies stage-specific modules as well as the branching cell-fate dynamics in human pluripotent stem cells with higher reconstruction accuracy than state-of-the-art methods. More importantly, GeoBridge supports single-cell fate navigation in multi-target hematopoietic lineages, allowing neutrophil-biased cells to be virtually guided toward mast-cell fates along biologically plausible paths. Together, GeoBridge establishes a principled pullback-geometric method that transforms sparse single-cell measurements into a continuous, controllable landscape for the reconstruction, navigation and manipulation of cellular state transitions.
- Open Access
- 25 June 2026
- Published by Wiley in Advanced Materials
- Vol. 38 (43) , e73876
A Biomimetic Single‐Atom Nanozyme With a Substrate Pocket for Accurate and Continuous Sweat Glucose Monitoring
Wearable non‐invasive glucose sensors are revolutionizing diabetic management by enabling real‐time monitoring and personalized care. However, their widespread adoption has been limited by the intrinsic instability and high cost of natural glucose oxidase (GOx). Although nanozymes offer a more robust alternative, their lack of substrate specificity is a major challenge. Herein, a manganese‐based tannic acid single‐atom nanozyme (TA‐Mn SA) with a dual biomimetic architecture was developed. It features atomically dispersed Mn catalytic sites embedded within a tannic acid‐derived molecular pocket tailored for selective glucose recognition. This engineered enzyme mimic exhibits exceptional catalytic efficiency and substrate selectivity. The origin of this selectivity and the catalytic cycle were elucidated through both experimental data and density functional theory calculations. When integrated into a wearable sweat sensor, the TA‐Mn SA nanozyme reliably tracked blood glucose levels in human trials, capturing personalized metabolic dynamics during fasting, feeding, and exercise. This design is inherently compatible with scalable, low‐cost manufacturing, paving the way for a new era of accessible continuous glucose monitoring.
- 25 June 2026
- Published by Wiley in Advanced Materials
- Vol. 38 (43) , e73876
Open AccessA Biomimetic Single‐Atom Nanozyme With a Substrate Pocket for Accurate and Continuous Sweat Glucose Monitoring
Wearable non‐invasive glucose sensors are revolutionizing diabetic management by enabling real‐time monitoring and personalized care. However, their widespread adoption has been limited by the intrinsic instability and high cost of natural glucose oxidase (GOx). Although nanozymes offer a more robust alternative, their lack of substrate specificity is a major challenge. Herein, a manganese‐based tannic acid single‐atom nanozyme (TA‐Mn SA) with a dual biomimetic architecture was developed. It features atomically dispersed Mn catalytic sites embedded within a tannic acid‐derived molecular pocket tailored for selective glucose recognition. This engineered enzyme mimic exhibits exceptional catalytic efficiency and substrate selectivity. The origin of this selectivity and the catalytic cycle were elucidated through both experimental data and density functional theory calculations. When integrated into a wearable sweat sensor, the TA‐Mn SA nanozyme reliably tracked blood glucose levels in human trials, capturing personalized metabolic dynamics during fasting, feeding, and exercise. This design is inherently compatible with scalable, low‐cost manufacturing, paving the way for a new era of accessible continuous glucose monitoring.
- 25 May 2026
- Published by Springer Nature in Nature Machine Intelligence
- Vol. 8 (6) , 997-1011
A large-scale unified deep learning model for peptide mass spectrum interpretation trained on multimodal data
- 25 May 2026
- Published by Springer Nature in Nature Machine Intelligence
- Vol. 8 (6) , 997-1011
A large-scale unified deep learning model for peptide mass spectrum interpretation trained on multimodal data
- Open Access
- 7 March 2026
- Published by Oxford University Press (OUP) in Genomics, Proteomics & Bioinformatics
The 2025 Westlake Autumn Symposium for Al Proteomics and Virtual Cell
- 7 March 2026
- Published by Oxford University Press (OUP) in Genomics, Proteomics & Bioinformatics
Open AccessThe 2025 Westlake Autumn Symposium for Al Proteomics and Virtual Cell
- Open Access
Accelerate the Highly Efficient Development of mRNA Vaccines Through Advanced Computational Methods
MRNA medicine is an emerging therapeutic approach that utilizes messenger RNA to synthesize functional proteins directly within target cells. This technology offers notable advantages including rapid development cycles, diverse therapeutic applications, and adaptable platform design for various diseases. However, mRNA therapeutic development faces substantial challenges, particularly in determining optimal mRNA sequences and developing effective delivery systems that ensure stability and achieve precise delivery. Current development processes often involve extensive experimental screening, highlighting the need for more efficient computational approaches. This review first introduces fundamental concepts in the mRNA vaccine field and systematically analyzes the roles and limitations of computational tools in advancing mRNA vaccine development across three key areas: sequence optimization, modification strategies, and delivery system optimization. Finally, we present the current application status of mRNA vaccines and discuss future prospects, highlighting emerging computational opportunities that may shape next‐generation mRNA vaccine development. This review spans the entire mRNA vaccine development pipeline, providing a foundational resource for researchers and facilitating technological advancement in this rapidly evolving field.
Open AccessAccelerate the Highly Efficient Development of mRNA Vaccines Through Advanced Computational Methods
MRNA medicine is an emerging therapeutic approach that utilizes messenger RNA to synthesize functional proteins directly within target cells. This technology offers notable advantages including rapid development cycles, diverse therapeutic applications, and adaptable platform design for various diseases. However, mRNA therapeutic development faces substantial challenges, particularly in determining optimal mRNA sequences and developing effective delivery systems that ensure stability and achieve precise delivery. Current development processes often involve extensive experimental screening, highlighting the need for more efficient computational approaches. This review first introduces fundamental concepts in the mRNA vaccine field and systematically analyzes the roles and limitations of computational tools in advancing mRNA vaccine development across three key areas: sequence optimization, modification strategies, and delivery system optimization. Finally, we present the current application status of mRNA vaccines and discuss future prospects, highlighting emerging computational opportunities that may shape next‐generation mRNA vaccine development. This review spans the entire mRNA vaccine development pipeline, providing a foundational resource for researchers and facilitating technological advancement in this rapidly evolving field.
- 29 August 2025
- Published by Wiley in Proteomics
- Vol. 25 (24) , 22-29
TDEase: An Open‐Source Data Visualization Software Framework for Targeted Proteoform Characterization by Top‐Down Proteomics
Top‐down proteomics (TDP) is a powerful approach for characterizing intact protein molecules and their diverse proteoforms. Despite recent advances, current TDP software tools often suffer from fragmented workflows, steep learning curves for non‐experts, or limited interactive visualization capabilities. To address these challenges, we introduce TDEase, an integrated analytical framework designed to streamline and enhance TDP data interpretation, with a current focus on integration with the TopPIC suite package for targeted proteoform characterization. TDEase features a modular architecture comprising TDPipe, a multi‐process data processing engine, and TDVis, an interactive web‐based visualization module. TDPipe automates the execution of mainstream TDP analysis algorithms through a user‐configurable pipeline, ensuring seamless and reproducible data processing. The TDVis module then transforms these results into dynamic, interactive dashboards, enabling multidimensional data exploration, including feature maps and PTM analysis. An alternative version, TDVisWeb, is also available for visualizing the results on an internet server or intranet workstation at institutional core facilities. We demonstrated the software capabilities in proteoform identification and comparative analysis using published histone datasets. TDEase is built with Python and open‐source, allowing future improvements and incorporation of more data types as the TDP community develops new software. Source code is available at https://github.com/Computational‐TDMS/TDEase .
- 29 August 2025
- Published by Wiley in Proteomics
- Vol. 25 (24) , 22-29
TDEase: An Open‐Source Data Visualization Software Framework for Targeted Proteoform Characterization by Top‐Down Proteomics
Top‐down proteomics (TDP) is a powerful approach for characterizing intact protein molecules and their diverse proteoforms. Despite recent advances, current TDP software tools often suffer from fragmented workflows, steep learning curves for non‐experts, or limited interactive visualization capabilities. To address these challenges, we introduce TDEase, an integrated analytical framework designed to streamline and enhance TDP data interpretation, with a current focus on integration with the TopPIC suite package for targeted proteoform characterization. TDEase features a modular architecture comprising TDPipe, a multi‐process data processing engine, and TDVis, an interactive web‐based visualization module. TDPipe automates the execution of mainstream TDP analysis algorithms through a user‐configurable pipeline, ensuring seamless and reproducible data processing. The TDVis module then transforms these results into dynamic, interactive dashboards, enabling multidimensional data exploration, including feature maps and PTM analysis. An alternative version, TDVisWeb, is also available for visualizing the results on an internet server or intranet workstation at institutional core facilities. We demonstrated the software capabilities in proteoform identification and comparative analysis using published histone datasets. TDEase is built with Python and open‐source, allowing future improvements and incorporation of more data types as the TDP community develops new software. Source code is available at https://github.com/Computational‐TDMS/TDEase .
- Open Access
Discovery of RNA‐Targeting Small Molecules: Challenges and Future Directions
RNA‐targeting small molecules represent a transformative frontier in drug discovery, offering novel therapeutic avenues for diseases traditionally deemed undruggable. This review explores the latest advancements in the development of RNA‐binding small molecules, focusing on the current obstacles and promising avenues for future research. We highlight innovations in RNA structure determination, including X‐ray crystallography, nuclear magnetic resonance spectroscopy, and cryo‐electron microscopy, which provide the foundation for rational drug design. The role of computational approaches, such as deep learning and molecular docking, is emphasized for enhancing RNA structure prediction and ligand screening efficiency. Additionally, we discuss the utility of focused libraries, DNA‐encoded libraries, and small‐molecule microarrays in identifying bioactive ligands, alongside the potential of fragment‐based drug discovery for exploring chemical space. Emerging strategies, such as RNA degraders and modulators of RNA–protein interactions, are reviewed for their therapeutic promise. Specifically, we underscore the pivotal role of artificial intelligence and machine learning in accelerating discovery and optimizing RNA‐targeted therapeutics. By synthesizing these advancements, this review aims to inspire further research and collaboration, unlocking the full potential of RNA‐targeting small molecules to revolutionize treatment paradigms for a wide range of diseases.
Open AccessDiscovery of RNA‐Targeting Small Molecules: Challenges and Future Directions
RNA‐targeting small molecules represent a transformative frontier in drug discovery, offering novel therapeutic avenues for diseases traditionally deemed undruggable. This review explores the latest advancements in the development of RNA‐binding small molecules, focusing on the current obstacles and promising avenues for future research. We highlight innovations in RNA structure determination, including X‐ray crystallography, nuclear magnetic resonance spectroscopy, and cryo‐electron microscopy, which provide the foundation for rational drug design. The role of computational approaches, such as deep learning and molecular docking, is emphasized for enhancing RNA structure prediction and ligand screening efficiency. Additionally, we discuss the utility of focused libraries, DNA‐encoded libraries, and small‐molecule microarrays in identifying bioactive ligands, alongside the potential of fragment‐based drug discovery for exploring chemical space. Emerging strategies, such as RNA degraders and modulators of RNA–protein interactions, are reviewed for their therapeutic promise. Specifically, we underscore the pivotal role of artificial intelligence and machine learning in accelerating discovery and optimizing RNA‐targeted therapeutics. By synthesizing these advancements, this review aims to inspire further research and collaboration, unlocking the full potential of RNA‐targeting small molecules to revolutionize treatment paradigms for a wide range of diseases.