Statistical and Computational Methods on GWAS and post-GWAS Analysis to Identify Genetic Basis of Intracranial Aneurysm

Saccular intracranial aneurysms (IAs) are balloon-like dilations of the intracranial arterial wall and affect approximately 2% of the population. Approximately 1% of these lesions rupture per year, resulting in subarachnoid hemorrhage (SAH). Substantial genetic contribution to both IAs (four-fold sibling recurrence risk) and SAHs (heritability of 41%) has been reported.;In this thesis, we first conducted a genome-wide association meta-analysis using 7,267 IA cases and 22,994 controls to identify IA risk loci through imputation using 1000 genomes phase 1 reference haplotypes. We analyzed a new case-control cohort from United Kingdom to follow up the finding at discovery stage using the targeted genotyping. We replicated 7 previously identified IA risk loci by our newly included GWAS cohorts (UGS, FIA and JP), as well as by the replication cohort from UK. 8 previously unknown risk loci of IA were identified from discovery stage and 6 of them were replicated in UK.;To identify genes underlying the association, we used a publicly available eQTL analysis data and tested the association between eQTLs and IA risk loci using a colocalization Bayesian test. We confirmed causal genes for some known loci and found potential causal genes for novel loci by comparing gene expression levels in different tissues and the colocalization of eQTLs with risk SNPs in IA. Chromatin states that were inferred using epigenomic data with a multivariate hidden Markov model were also incorporated to fine map the risk loci.;The differentiated association test that allows different effect sizes (and direction) between clinical subtypes was performed to explore the genetic correlations between IA and other phenotypes. We focused on genomic characteristics of the 15 IA risk loci and searched for associated SNPs for other phenotypes in the same locus of IA. 5 of the IA risk loci confer opposite risk to atherosclerosis-related phenotypes, including the carotid intima-media thickness (IMT), the low-density lipoprotein (LDL) cholesterol and the large vessel stroke, as well as coronary heart disease. Summary statistics from 16 traits were exploited. The genetic correlations for pair-wise comparisons among the 16 traits and IA were evaluated using LD score regression. We found negative correlations between IA and LDL and between IA and Type 2 Diabetes.