Date of Award

Winter 12-15-2022

Author's School

Graduate School of Arts and Sciences

Author's Department

Biology & Biomedical Sciences (Computational & Systems Biology)

Degree Name

Doctor of Philosophy (PhD)

Degree Type



Our understanding of gene regulation needs to be generalizable as well as specific. A generalizable understanding enables us to transfer our knowledge of one gene to another, and the specificity allows us to understand the precise regulation of gene expression through development. One framework that confers generalizability and specificity is the hierarchical and modular model of gene regulation. To test whether gene regulation is hierarchical and modular requires methods to systematically analyze how different factors collectively control gene expression. This thesis describes the development of two functional genomics methods at single-cell resolution. These methods systematically examined how cellular contexts, chromatin environment, and local regulatory sequence collectively control expression mean and noise. First, scMPRA measures cell-type specific expression of a library of core promoters in K562 and HEK 293 cell lines. scMPRA can also be applied to a complex tissue and performed MPRA ex vivo. Both general principles of gene expression and cell-type specific variant effects are found in the newborn mouse retina. scMPRA also measured the cell substate effect on expression mean, I found that cell substate has a large and general effect on expression mean for core promoters. I also deconvolved the extrinsic and intrinsic portion of expression noise, we found that developmental core promoters have larger extrinsic noise. Second, scTRIP measures the chromatin environment effect on expression noise. We found expression noise can be partially explained by expression noise. We also found the expression noise that is independent of expression mean is correlated with specific chromatin marks and transcription factor binding sites. Moreover, we identified the oscillation between cell substates as a major source of extrinsic noise regardless of the chromatin environment. Using all the information, we trained a logistic regression model with high accuracy. These observations and methods provide a framework to further explore the hierarchical and modular nature of gene regulation.


English (en)

Chair and Committee

Barak A Cohen

Committee Members

Zachary Pincus