Within our lab, we aim to formulate a range of biomedical challenges into AI-driven problems by designing powerful deep learning architectures and developing efficient computational frameworks. Our work is mostly problem-driven and methodology focused, and we benefit greatly from close collaborations with experimental labs.
1. Computational methods for regulatory analysis of transcription process.
Computational methods for regulatory analysis of the transcription process are among the core tools in understanding gene expression regulation. In recent years, the explosion of single cell and spatial multiomics data spanning genomics, transcriptomics, and epigenomics allows us to recognize more complex patterns and regulatory mechanisms in transcription processing. For example, prevalent heterogeneity of features such as splicing efficiency across cell types and contexts, the dynamic interplay between RNA-binding proteins and cis-regulatory elements. Based on available high-throughput sequencing data, we are interested in finding the biological mechanisms and driving factors of such patterns, and how to combine these factors into transcriptional regulatory analyses.
2. Artifical intelligence driven antibody design.
Protein language models represent a powerful paradigm in artifical intelligence, capable of learning latent sequence-structure-function relationships directly from amino acid sequence data. However, generating antigen-specific antibodies without prior templates remains a major challenge due to the highly diverse and dynamic nature of antibody-antigen recognition interfaces. We aim to develop and fine-tune protein language models to learn the complex pairing rules between human antibody variable heavy and light chains and their binding specificities toward diverse viral antigens. Furthermore, we seek to interpret the sequence determinants of binding specificity, neutralization breadth, and epitope diversity using attention-based and explainable AI methods, bridging generative antibody design with mechanistic understanding of immune recognition.
3. Other directions.
We are broadly interested in the topics of single-cell and spatial biology, tumor microenvironment,
digital pathology.