In partial fulfillment of the requirements for the degree of
Doctor of Philosophy in Bioinformatics
in the School of Biological Sciences

Bridget Neary 

Defends her thesis:

Biological Insights from Predictors of Drug Response in Cancer Patients

Monday, December 1, 2025 at 1:00 PM

Engineered Biosystems Building (EBB),

CHOA seminar room EBB 1005

Meeting Link: Bridget Neary Thesis Defense | Microsoft Teams 

Thesis Advisor

Dr. Peng Qiu 
Department of Biomedical Engineering
Georgia Institute of Technology

Committee Members

Dr. Greg Gibson
School of Biological Sciences
Georgia Institute of Technology

Dr. Saurabh Sinha
Department of Biomedical Engineering
Georgia Institute of Technology

Dr. Manoj Bhasin
Department of Pediatrics
Emory University

Dr. Kevin Bunting
Department of Pediatrics
Emory University

Abstract

Cancer therapies show considerable variation in efficacy, even among histologically similar cancers, and while precision medicine has made great strides in addressing this fundamental challenge, current advances still only benefit a minority of patients. Extending the reach of precision medicine will require the development of clinically deployable methods of predicting therapeutic outcomes and the expansion of our mechanistic understanding of variability of therapeutic efficacy. This thesis addresses these needs through three complementary approaches. First, we identified clinically relevant transcriptomic and methylomic single- and multi-gene biomarkers of treatment response in pre-treatment primary tumors. Next, to elucidate cellular mechanisms of variability in therapeutic variability, we investigated common transcriptional patterns across patients in each cancer that were predictive of drug response. Enrichment analysis of multi-gene biomarkers and response-associated transcriptional patterns suggested upstream regulators potentially involved in cellular drug effects, and analysis of core gene sets linked known chromosomal aberrations to patient differences in therapeutic response. Finally, we developed a multi-drug response prediction model that predicted clinical drug efficacy, adapting a biologically informed neural network structured around the Gene Ontology hierarchy for model interpretability and using transfer learning to leverage large preclinical datasets to compensate for insufficient patient sample sizes. Network layer importance analysis implicated known and novel biological processes involved in response to specific drugs and in multi-drug resistance. Together, these results provide new tools for predicting cancer treatment outcomes and new insights into cellular mechanisms of drug response. This work demonstrates how identification and development of predictors of patient treatment outcomes can be leveraged for additional understanding of the underlying biology that is essential for continued development and clinical implementation of precision medicine tools with broader clinical reach.