My research focuses on identifying genetic variants, genes, and molecular biomarkers associated with cancer risk, and use these target to guide risk prediction and drug repurposing for cancer prevention and intervention. These work involves the integration of both bulk and single-cell multi-omics data of blood and cancer-relevant tissues, with genome-wide association study (GWAS) data of cancers. The work I expect to be done with my USOAR student aims to identify cell-type-specific epigenetic marks and genes associated with pancreatic cancer risk.
Pancreatic cancer is the third leading causal of cancer death in the U.S., highlighting the critical needs for effective prevention and treatment strategies. Targeting causal genes and biological mechanisms revealed by human genetic studies offers a promising avenue, as drug mechanisms supported by genetic evidence is ~2.6 time more likely to succeed that those without. However, although genome-wide association studies have identified 33 susceptibility loci for pancreatic cancer, the causal genes and biological mechanisms for most of these loc remain unclear.
We and others have conducted transcriptome-wide association studies (TWAS) and DNA methylome-wide association studies (MeWAS), uncovering multiple gene and DNA methylation biomarkers associated with pancreatic cancer risk. However, these genes and biomarkers only account for a limited number of GWAS-identified loci. In addition, most of these studies used bulk epigenomic and transcriptomic data with limited resolution of genetically regulated epigenetic and gene expression changes. To address this, we plan to conduct a single-cell epigenome- and transcriptome-wide association study (scEWAS and scTWAS) by integrating single-cell multiome (ACAC-seq + RNA-seq) data of pancreas tissues with pancreatic cancer GWAS data. We will develop cell-type-specific prediction models for chromosome accessibility and gene expression using genetic variants. These models will be applied to pancreatic cancer GWAS data to assess the association of genetically predicted chromosome accessibility and gene expression with pancreatic cancer risk. Identified associations will then be evaluated for their heterogeneity across cell types, as well as there potential causality through colocalization and summary-data-based Mendelian randomization analyses. We will further explore the druggability of prioritized causal genes using drug-target databases and in vitro drug perturbation data.
We expect that the findings from this study will identify novel epigenetic biomarkers and genes for pancreatic cancer and point to drug repurposing candidate that warrant further investigation.
Students with programming experiences using Linux Shell/R/Python is preferred.
1. Knowledge in pancreatic cancer genetics and multi-omics data.
2. Hands-on experiences in analyzing genetic and multi-omics data.
3. Experiences in manuscript/grant writing.