Research Interests:

Virtual Cell Modeling and Disease Driver Discovery:

Disease initiation and progression are driven by functionally important genetic alterations and their interactions within complex molecular networks. Our group integrates genomic, transcriptomic, epigenomic, single-cell, and other multi-omics data with evolutionary modeling, artificial intelligence, and graph-based learning to construct virtual cell models that capture disease-associated molecular states and regulatory interactions. These models enable the systematic identification and prioritization of disease-driving mutations, genes, and gene modules, as well as the discovery of potential therapeutic targets. By distinguishing causal driver alterations from passenger events and reconstructing their effects across cellular and molecular networks, we aim to establish an interpretable computational framework for disease mechanism analysis, target discovery, and precision medicine.

Neoantigen Prediction and Immune Microenvironment Profiling:

Tumor-specific neoantigens generated by somatic mutations represent highly attractive targets for cancer immunotherapy. Our group develops artificial intelligence and bioinformatics methods to predict neoantigens with antigen processing, HLA–peptide binding, and T-cell recognition. We further integrate bulk and single-cell multi-omics data, immune-repertoire information, and TCR–antigen interaction resources to characterize the tumor immune microenvironment and investigate the relationships among tumor cells, antigen-presenting cells, T cells, and neoantigen responsiveness. These efforts aim to identify antigenic targets that are not only computationally predicted, but also biologically presented and immunologically recognizable in individual patients.

Immune Receptor Evolution and Biotherapeutic Design:

Building on the discovery of cancer-driving alterations and the characterization of their immune recognition potential, our group develops engineered immune receptors and targeted biologics that translate antigen recognition into selective antitumor activity. We employ directed-evolution and protein-engineering strategies to optimize T-cell receptors (TCRs) for enhanced affinity, specificity, stability, and functional recognition of tumor-associated antigens and neoantigens, while minimizing off-target reactivity. In parallel, we design T-cell engagers (TCEs) that selectively recruit and activate T cells against antigen-expressing tumor cells. Together with personalized neoantigen vaccine development, these efforts establish an integrated pipeline from computational target discovery and immune-receptor optimization to therapeutic molecule design and translational evaluation. Notably, personalized cancer vaccines designed using our computational algorithms have advanced to the clinical-trial stage.