Computational approaches to study molecular signatures of brain aging and neurodegeneration

The molecular heterogeneity of aging and neurodegeneration often makes it difficult to identify the specific drivers of disease progression, especially when analyzed at the bulk-tissue level. While single-cell analysis methods can overcome bulk-tissue limitations, they typically examine individual aspects of cellular behavior in isolation, missing important relationships that can only be detected when multiple analytical layers are integrated.

This thesis addresses these limitations through an integrated computational framework combining differential expression analysis, pathway analysis, network analysis, and cell-cell communication analysis. We applied this framework to study Alzheimer’s disease (AD) and Parkinson’s disease (PD) using human datasets alongside mouse and zebrafish models. While the framework identified common neurodegenerative signatures across both diseases, our work focused specifically on identifying sex-specific differences in AD and validating findings across species. Using AD as a primary case study, the analysis of 2.3 million single-cell transcriptomes from human AD brains revealed marked sex-specific molecular patterns. In male AD patients, astrocytes showed increased activity in apoptotic and cell death pathways, while female astrocytes displayed distinct changes in Wnt signaling and cell-cycle regulation. We identified specific genes and pathways showing sex-specific and sex-dimorphic changes, revealing how cellular interactions may contribute differently to AD pathology in men and women. The integrated analytical approach revealed patterns of intercellular signaling that conventional single-layer analyses failed to detect.

These findings contribute to our mechanistic understanding of aging and neurodegeneration while providing a systematic and reproducible computational framework for systems-level omics analyses of neurodegenerative diseases.