| Most common diseases have substantial heritable components but are characterized by complex inheritance patterns implicating numerous genetic and environmental factors. A longstanding goal of human genetics research is to delineate the genetic architecture of these traits - the number, frequencies, and effect sizes of disease-causing alleles - to inform mapping studies, elucidate mechanisms of disease, and guide development of targeted clinical therapies and diagnostics. Although vast empirical genetic data has now been collected for common diseases, different and contradictory hypotheses have been advocated about features of genetic architecture (e.g., the contribution of rare vs. common variants). Here, we present a framework which combines multiple empirical datasets and simulation studies to enable systematic testing of hypotheses about both global and locus-specific complex trait architecture. We apply this to type 2 diabetes (T2D).;For T2D, we find that extreme models of global genetic architecture are excluded (e.g., models where T2D is a collection of rare Mendelian diseases), but a wide range of models remain consistent with epidemiology, linkage, and genome-wide association studies (GWAS). Simulations predict that ongoing sequencing and genotyping studies (in tens of thousands of individuals) will further constrain architecture, but that very large sample sizes (e.g., >250K unselected individuals) will be required to localize most T2D heritability.;To characterize allelic architecture at individual T2D loci, we develop haplotype-based methods to integrate data from GWAS and low-pass sequencing of thousands of T2D cases and controls. We find varied architectures plausible at each locus. At some loci, the most likely model implicates common causal variation (chr9p21, TCF7L2, KCNJ11, HNF1B). At others, there is evidence for common variants of weak effect alongside independent low-frequency variants of larger effect (CCND2, KCNQ1) or a burden of very rare protein-coding, disease-associated mutations (PPARG). Finally, at several loci, further genetic and/or experimental interrogation is required to determine whether causal alleles are common, rare, or both (HMGA2, IGF2BP2).;In this thesis, we have integrated diverse datasets to better understand the genetic and biological architecture of T2D. This work informs future genetic and experimental studies of T2D, and provides methods for hypothesis testing that are broadly applicable to many complex traits. |
