Date of Award

5-12-2023

Document Type

Dissertation

Degree Name

Doctor of Philosophy (PhD)

Department

Mathematics and Statistics

First Advisor

Wei Zhang

Abstract

Identifying gene–gene and gene–environment interaction is complicated and gainsaying most especially because of high multifactor dimensions involve in the analysis of such combination discrediting the functionality of parametric statistical method like Logistic Regression. [22] developed a multi-factor-dimensionality reduction (MDR) method for detecting and characterizing high-order gene-gene and gene-environment interactions in case-control and discordant-sib-pair studies with relatively small samples, which is inspired by the combinatorial partitioning method of [18]. [16] proposed Generalized MDR (GMDR) framework based on the score of a generalized linear model which allows adjustment of covariates, provides a unified framework for handling both dichotomous and quantitative phenotypes. However, there are limitations for MDR and GMDR. First, the MDR and GMDR can handle diverse types of phenotype and various study designs for identification interactions between/among discrete variables. Second, MDR and GMDR can’t be used to identify SNPs and combinations of SNPs that relate to the survival phenotype in many cancer studies, especially under depend censoring. KM-MDR and Cox-MDR were proposed for detecting gene–gene interactions associated with survival time based on Kaplan-Meier estimator and Cox proportional hazard model, which can be considered as extension of MDR and GMDR, respectively. However, all these methods rely on the independent censoring assumption: survival time and censoring time need to be statistically independent. This assumption is easily violated when patients drop out from the study due to the worsening of his/her health condition or patients are removed for transplantation. In this case, standard survival techniques, like Kaplan-Meier estimator and Cox proportional hazard model, give biased results. Copula-graphic estimator was used instead of Kaplan-Meier estimator or cox proportional hazard model to correct for bias due to dependent censoring. In the first part of the dissertation, we proposed a model based Generalized Additive Semiparametric method to improve on the downside enumerated above for identification interactions between/among discrete variables. This method will allow for covariate adjustment, reduces the computational problems experienced in MDR and incorporate linear or non-linear interaction of genetic factors with environmental factors. In the second part of the dissertation, copula-graphic based estimators are used for detecting gene–gene and gene-environment interactions associated with survival time, which can effectively adjust for dependent censoring and yield more reliable results. We then conducted an extensive series of Monte Carlo simulations to compare the performances of these methods in terms of predicting accuracy and power considering different scenarios by varying interaction type and censoring type. We also compared the performance of these methods when applied for a real data.

Download

Included in

Biostatistics Commons

Share

COinS