Ph.D. Defense: Jasmine Yang
Ph.D. Thesis Defense
Wednesday, May 1st, 2019
125 Hanes Hall
Liuqing (Jasmine) Yang
Statistical Methods for Deconvolution in Cancer Genomics
(Under the direction of J. S. Marron and Hongtu Zhu)
The heterogeneity within a bulk tumor tissue, referred as intratumor heterogeneity, becomes a prevalent confounding factor to tumor genomic studies. Analysis on genomic profiles from heterogeneous tumor samples can potentially lead to false positive differential expression conclusions, and even influence patients’ clinical outcomes and therapeutic responses. To address the intratumor heterogeneity issue, in this dissertation we first develop a Fast Tumor Deconvolution tool to separate the pure tumor signals from tumor-nontumor mixtures in an efficient way. Assuming a linear combination of the abundance of the mixing components and the availability of reference information for the non-tumor component, our semi-parametric regression-based model can quickly provide estimates for the tumor proportion in a mixture, as well as output the tumor specific genomic profile. Then the method is extended to deconvolve heterogeneous tumor samples with more than two subcomponents, facilitated by a new scheme for gene selection. Both simulation and real data application studies are performed to demonstrate the effectiveness of these methods.