Profiling the immune system in human diseases. We use a variety of statistical and algorithmic methods to analyze multi-omics data from immunological studies, including data from CyTOF, RNA-sequencing, single-cell sequencing, proteomics (Olink), antibody profiling, and high-throughput T cell assay experiments. We collaborate with multiple experimental groups to characterize the immune system in human diseases. We are particularly interested in identifying key cellular or molecular targets that can be used as drug targets, as well as developing machine-learning models to predict the outcomes of patients. The ongoing projects include:
- Longitudinal study of naturally acquired immunity in malaria patients.
- Deep immune profiling of COVID19 patients.
- Mining massive public data to discover modulators of ACE2 and TMPRSS2 expression.
- System immunology to characterize the impact of latent herpes virus.
- Proteomics profiling of cerebrospinal fluid from patients with HIV-associated dementia.
Developing methods for analyzing immunological data. The immune system is highly complex, involving the interaction of a variety of genes, proteins, and cells. Using high-throughput technologies such as mass cytometry (CyTOF), single-cell RNA-sequencing, and multiplex cytokine assays, researchers can now profile the immune system in great detail. The lab is interested in developing novel computational methods for analyzing multi-omics data from immunological studies. We have developed machine-learning methods for predicting clinical outcomes using immune profiling data, including CytoDx and deep learning-based models. We have also developed a computational tool named MetaCyto to enable automated meta-analysis of cytometry datasets, including data from both conventional flow and CyTOF cytometry data. MetaCyto is adopted by the 10,000 immunome project to characterize the immune cell populations in healthy individuals.
Data-driven drug discovery to improve cancer immunotherapy. CD8+ T cells play critical roles in cell-mediated immunity. Upon antigen recognition and proper co-stimulatory signals, CD8+ T cells will carry out a variety of effector functions. However, CD8+ T cells often enter suppressed states in tumors or chronic viral infections, preventing T cells from eliminating the tumor cells. Activating the suppressed T cells has shown to be an effective way of treating cancer. The lab is interested in using in-silico drug discovery methods to identify compounds that can reverse the molecular changes underlying the suppressed T cells, and test the drugs in mouse tumor models.