Research

Our lab is focused on the development of computational and statistical methods for extracting meaningful biological and translational information from data. The aim is to incorporate structural, domain, or side information to increase the analysis’s reliability. Our research sits at the intersection of Statistics, Machine Learning, Bioinformatics, Genomics, and Pathology. The lab encourages and promotes collaboration between biomedical and quantitative investigators and between senior and young researchers, including students and postdoctoral trainees.

Our current research interests include the following.

Single-cell and Spatial Multi-modality Data Inference

Single-cell and spatial omics data quantify different aspects of individual cells or a small number of cells, allowing us to explore high-resolution biological heterogeneity. We focus on developing methods to reveal the complex heterogeneity structure among cells and tissue locations and to integrate information in multi-modality data. We aim to reveal complex biological processes and assist the development of possible translational solutions.

Integrative Statistics and Artificial Intelligence (AI) Modeling

Classic statistical models and AI models are two commonly used modeling strategies in data science. Both strategies have their pros and cons. Classic statistical models require fewer sample sizes and provide more interpretable results. Nevertheless, their performance is poor when complex data structure violates the pre-assumed model. In contrast, AI models automatically explore complex data structures. Nevertheless, if the sample size is too small, AI models often yield less reliable results. Besides, AI models have weaker interpretability because their primary goal is prediction (or optimization). We focus on developing algorithms to integrate two types of modeling to amplify their strengths and increase the analysis accuracy and interpretability.