SEMI-PARAMETRIC BAYESIAN FUNCTIONAL MAPPING WITH IRREGULAR SPARSE LONGITUDINAL DATA
Open Access
- Graduate Program:
- Statistics
- Degree:
- Doctor of Philosophy
- Document Type:
- Dissertation
- Date of Defense:
- April 15, 2011
- Committee Members:
- Rongling Wu, Dissertation Advisor
- Rongling Wu, Committee Chair
- Runze Li, Committee Chair
- Vernon Michael Chinchilli, Committee Member
- Debashis Ghosh, Committee Member
- Peter Cm Molenaar, Committee Member
- Keywords:
- Longitudinal data
- functional mapping
- MCMC
- Abstract:
- Genome-wide Association Studies, popularly known as GWAS, are playing a key role in understanding the genetic patterns of various traits and diseases. Despite the potential importance of GWAS, several limitations of such studies have been recognized in recent years. While most of the genetic traits and diseases are dynamic, most GWAS consider a single time point phenotypic measurement which results in reduced statistical power and practical usefulness. In this thesis, we have incorporated the functional aspect of the dynamic traits by considering repeated measurements over subject-specific time points. For that, we have developed semiparametric approach for joint modelling of the genotype-specific mean trajectories and the covariance matrix. Since for most of the biomedical experiments, it is of interest at what time point a particular event (death of a patient, for example) occurs, we also develop a framework for joint modelling of the longitudinal and survival data. Semi-parametric approach for joint modelling of the mean and covariance function for bivariate longitudinal trait has been proposed here too. All our proposed approaches have been validated by real data as well as extensive simulation studies.