UCLA

DNA methylation underpins a wide range of biological processes and disease states, yet significant challenges persist in its computational analysis and practical application. This dissertation presents advancements in three areas of DNA methylation research, including simulation method development and analysis in disease and aging. These advancements contribute to improved methodologies and a deeper understanding of the DNA methylome, paving the way for fundamental research and clinical applications.

In the first part, we introduce BSReadSim, a novel bisulfite sequencing simulator that addresses the limitations of existing tools, which often fail to capture the complexities of real-world data. By accurately integrating genetic variants, methylation profiles, and technical artifacts, BSReadSim can generate realistic synthetic datasets. It offers a powerful framework for designing experiments, developing computational methods, and benchmarking analytical pipelines in DNA methylation research, ultimately enhancing the rigor and reliability of epigenetic studies.

The second part of this dissertation pioneers the use of saliva DNA methylation for type 2 diabetes (T2D) biomarker discovery and risk assessment. Utilizing a cost-effective, two-step approach that combines Whole Genome Bisulfite Sequencing (WGBS) and Targeted Bisulfite Sequencing (TBS), we identified and characterized DNA methylation biomarkers associated with T2D. Importantly, we demonstrated that these epigenetic signatures in saliva are primarily intrinsic rather than driven by cell composition shifts, establishing saliva DNA methylation as a compelling non-invasive biomarker for T2D with significant potential for research and clinical diagnostics.

The third part of this work investigates the interplay between epigenetic aging, cell composition, and breast cancer risk in normal breast tissue. By analyzing 181 normal breast samples, we revealed systematic biases in existing epigenetic clocks, emphasizing the need for tissue-specific clocks to achieve accurate age prediction. Our findings demonstrated a clear link between epigenetic age acceleration and shifts in cell composition, particularly those associated with elevated breast cancer risk. Notably, we provided molecular evidence, for the first time, connecting estrogen exposure to accelerated epigenetic aging and increased cancer susceptibility. These insights highlight the potential of epigenetic clocks as powerful tools for cancer risk assessment and stratification.

Together, these studies deepen our understanding of DNA methylation's role in health and disease and highlight the transformative potential of innovative tools and methodologies in this field. By advancing computational methods, enabling biomarker discovery, and uncovering mechanistic insights, this thesis establishes a strong foundation for future research and clinical advancements in epigenetics. It paves the way for developing rigorous computational tools, innovative diagnostic approaches, and novel therapeutic strategies.