Data-adaptive SNP-set-based association tests of longitudinal traits
Genome-wide association studies (GWASs) have been largely limited to investigating traits with a single time measurement. However, many prospective cohort studies and electronic health record (EHR)-based cohorts with GWAS data have collected traits with repeated measurements across follow-up time. Effectively utilizing the information embedded in the time trajectory of measurements could greatly increase the power of association testing. Local association signal patterns across the whole genome are usually variable, complicated and unpredictable, which underscores the need for a data adaptive test capable of maintaining high statistical power across different genetic architecture and association signal patterns. Furthermore, complex diseases are usually affected by multiple variants in a gene and multiple genes in a biological pathway. In addition, traditional single SNP-based association testing has very limited statistical power for variants of low to rare minor allele frequency (MAF < 5% or 1%). A SNP-set-based association test, e.g., based on genes or pathways, could boost the statistical power by aggregating individual weak to moderate association signals across a region of interest. In this dissertation, I have developed such powerful data-adaptive tests that address these analytical challenges facing association testing between longitudinal traits and rare or common variants. I implemented extensive simulation studies to evaluate the performance of the proposed tests, and illustrated their applications in the Atherosclerosis Risk in Communities (ARIC) study. I also produced a software package with documentation to implement the proposed tests in high performance computing platform. In conclusion, this dissertation paves a new path in extending the traditional association tests to longitudinal traits, helps identify novel genes and explain the missing heritability in human complex diseases.
Yang, Yang, "Data-adaptive SNP-set-based association tests of longitudinal traits" (2015). Texas Medical Center Dissertations (via ProQuest). AAI10109675.