UC Irvine

Alzheimer’s disease and related neurodegenerative disorders involve progressive molecular and cellular alterations across diverse brain cell types, yet the regulatory mechanisms driving these changes remain incompletely understood. Single-cell transcriptomic and epigenomic technologies now allow detailed characterization of disease-associated cell states, but major challenges remain in defining the gene regulatory programs underlying disease progression and determining how these programs may be functionally perturbed. This dissertation addresses these challenges by integrating single-cell multiomic analysis with computational modeling of gene regulatory networks to identify disease-associated regulatory circuits and evaluate their roles in cellular state transitions.To define disease-associated regulatory programs, I first examined transcriptomic and epigenomic alterations in tauopathies, focusing on Pick’s disease and Alzheimer’s disease (Chapter 1). Comparative analyses revealed disease-enriched non-coding regulatory regions, altered predicted transcription factor binding, and enhancer-linked target genes across neuronal and glial populations. Fine-mapping of Alzheimer’s disease risk loci further identified enrichment in microglial enhancers and accessible regions in additional cell types. Functional relevance was supported through CRISPR-mediated excision of a predicted UBE3A enhancer, and the resulting regulatory predictions were made accessible through scROAD, an interactive resource for visualizing single-cell transcription factor occupancy and regulatory networks. Building on this regulatory framework, I developed compact, a computational approach for co-expression module perturbation analysis in single-cell transcriptomic data (Chapter 2). compact extends high-dimensional weighted gene co-expression network analysis by simulating knock-in, knock-down, or knock-out perturbations of network hub genes and propagating perturbation effects through gene network structure. Applications to human disease-associated microglia and single-cell perturbation datasets showed that perturbing disease-relevant modules can induce predicted phenotypic shifts and prioritize candidate regulatory drivers with therapeutic relevance. Finally, I extended this framework to Alzheimer’s disease gray and white matter, with emphasis on oligodendrocytes and other glial populations (Chapter 3). Initial single-nucleus transcriptomic analyses identified region- and compartment-specific cellular states and co-expression programs associated with myelination, inflammatory signaling, and glial-neuronal interactions. These findings provide a foundation for future integration of epigenomic profiling and in silico perturbation to examine whether oligodendrocyte regulatory modules can be reprogrammed toward disease-resilient states. Together, these studies move beyond descriptive single-cell disease atlases by characterizing neurodegeneration through regulatory modules whose structure, function, and perturbability can be systematically investigated. This dissertation provides a framework for interpretable, network-guided therapeutic hypothesis generation in neurodegenerative disease.