Virtual Cells for Precision Medicine
Modeling cellular states and responses across diverse biological and patient contexts to support precision medicine.
Artificial Intelligence & Computational Biology
Ziheng Duan
Postdoctoral researcher · Yale University
I develop AI methods to model biological systems and guide therapeutic discovery. My current interests include virtual cells for precision medicine, therapeutic intervention design, and closed-loop AI for science.
I earned my Ph.D. in Computer Science at UC Irvine, advised by Prof. Jing Zhang. My doctoral research developed graph learning methods for single-cell and spatial omics, from genomic regulation to multi-slice tissue atlases. I also work with Genesis Molecular AI on therapeutic target identification and immune cell atlas analysis.
Education
Ph.D. in Computer Science
UC Irvine · 2025
B.S. in Automation and Computer Science
Zhejiang University · 2020
Latest updates
News
Joining Prof. Mark Gerstein’s lab at Yale University as a postdoctoral researcher in November 2026.
Two papers accepted to NeurIPS 2026: “TargetSage: Identifying Therapeutic Target Genes with Interpretable and Robust LLM Reasoning” and “sMMC-22M: A Context-Aware Dataset and Benchmark for Single-Cell Spatial Transcriptomics.”
Presented AI-Driven Therapeutic Target Discovery and Cell-State Reprogramming at the SCORCH 2026 Fall Meeting in San Diego. Talk details ↗
Earned my Ph.D. in Computer Science from UC Irvine.
Our collaborative study, “Single-cell transcriptomic and chromatin dynamics of the human brain in PTSD,” was published in Nature. Read the paper ↗
Our collaborative study, “Single-cell genomics and regulatory networks for 388 human brains,” was published in Science. Read the paper ↗
01 / Research
Current Research Interests
Modeling cellular responses, designing therapeutic interventions, and learning from scientific feedback.
AI for Therapeutic Discovery
Identifying therapeutic targets and designing interventions to steer disease-associated cells toward healthier states.
Closed-Loop AI for Science
Exploring AI systems that connect hypothesis generation, experimental design, and learning from experimental feedback.
PhD Research · UC Irvine
From Cells to Atlases
Graph Learning for Single-Cell and Spatial Omics
My doctoral research developed graph learning methods for single-cell and spatial omics across biological scales—from chromatin organization and disease-associated gene networks to cellular microenvironments and multi-slice tissue atlases.
02 / Publications
Selected work
Single-cell transcriptomic and chromatin dynamics of the human brain in PTSD
Nature · 2025
Mapping cell-type-specific gene expression and chromatin changes in the human brain in PTSD.
MUSE: A Multi-slice Joint Analysis Method for Spatial Transcriptomics Experiments
CIKM · 2025
Jointly analyzing multiple spatial transcriptomics slices to identify spatial domains.
iMIRACLE: An Iterative Multi-View Graph Neural Network to Model Intercellular Gene Regulation From Spatial Transcriptomic Data
CIKM · 2024
Connecting cell–cell communication with intracellular gene regulation in spatial transcriptomics.
Impeller: a path-based heterogeneous graph learning method for spatial transcriptomic data imputation
Bioinformatics · 2024
Imputing spatial gene expression by combining spatial proximity and expression similarity.
Single-cell genomics and regulatory networks for 388 human brains
Science · 2024
Mapping cell-type-specific gene regulation across 388 human brains.
scENCORE: leveraging single-cell epigenetic data to predict chromatin conformation using graph embedding
Briefings in Bioinformatics · 2024
Reconstructing chromatin compartments from single-cell epigenetic data using graph learning.
iHerd: an integrative hierarchical graph representation learning framework to quantify network changes and prioritize risk genes in disease
PLOS Computational Biology · 2023
Quantifying gene network rewiring to prioritize disease-associated risk genes.
03 / Experience
Industry experience
Apr 2026 – present
Consultant
Genesis Molecular AI Conducting computational analyses of immune cell atlases.
Jun – Sep 2025
Machine Learning Research Intern
Genesis Molecular AI Developed machine learning methods integrating multi-omics data and biomedical knowledge to identify therapeutic targets.
04 / Talks
Talks
2026
From Immune Atlases to Therapeutic Reprogramming
Talk · Genesis Molecular AI · September 22, 2026
2026
AI-Driven Therapeutic Target Discovery and Cell-State Reprogramming
Talk · SCORCH 2026 Fall Meeting · September 15, 2026 · San Diego, California
2024
scENCORE: Reconstructing Chromatin Compartments from Single-Cell Epigenetic Data
Talk · 13th Annual Symposium, Southern California Systems Biology · May 2024
2023
iHerd: Identifying Key Driver Genes through Changes in Gene Networks
Talk · SCORCH Consortium Meeting · December 5, 2023
05 / Service
Academic community
Google Summer of Code
Mentor
- 2026 · StaR — A Stability-Aware Representation Learning Framework for Spatial Domain Identification
- 2025 · RAG-ST — Retrieval-Augmented Generation for Spatial Transcriptomics
- 2024 · BenchmarkST — Spatial Transcriptomics Gene Imputation Benchmarking
Teaching
Reader · UC Irvine
- ICS 178 · Machine Learning and Data Mining
Winter & Fall 2021 - ICS 6D · Discrete Mathematics for Computer Science
Spring 2022
Reviewing (selected)
Conferences
ISMB · Learning on Graphs (LoG) · IEEE BIBM
Journals
Nature Communications · Genome Medicine · BMC Medicine · IEEE TPAMI · IEEE TKDE · Pattern Recognition · Neural Networks