Statistical Methods For Genetic Variants Detection With Epigenomic Information

Genome-wide association studies (GWAS) have successfully identified thousands of genetic variants contributing to disease and other phenotypes. However, significant obstacles hamper our ability to elucidate causal variants, identify genes affected by causal variants, and characterize the mechanisms by which genotypes influence phenotypes. The increasing availability of genome-wide functional annotation data provides unique opportunities to incorporate prior information into the analysis of GWAS to better understand the impact of variants on disease etiology. Regulatory genomic information has been recognized as a potential source that can improve the detection and biological interpretation of single-nucleotide polymorphisms (SNPs) in GWAS. Although there have been many advances in incorporating prior information into the prioritization of trait-associated variants in GWAS, functional annotation data has played a secondary role in the joint analysis of GWAS and molecular (i.e., expression) quantitative trait loci (eQTL) data in assessing evidence of association. Moreover, current methodologies that aim to integrate such annotation information focus mainly on fine-mapping and overlook the importance of its usage in earlier stages of GWAS analysis. Equally important, there is a lack of development in proper statistical frameworks that can perform selection of annotations and SNPs jointly. To address these shortcomings, we develop two statistical models: iFunMed and GRAD. iFunMed is a novel mediation framework to integrate GWAS and eQTL data with the utilization of publicly available functional annotation data. iFunMed extends the scope of standard mediation analysis by incorporating information from multiple genetic variants at a time and leveraging variant-level summary statistics. GRAD integrates high-dimensional auxiliary information into high-dimensional regression. This method allows annotation information to assist the detection of important genetic variants while identifying relevant annotation simultaneously. We provide an upper bound for the estimation error of the SNP effect sizes to gain insights on what factors affect estimation accuracy. For iFunMed, data-driven computational experiments convey how informative annotations improve SNP selection performance while emphasizing the robustness of the model to non-informative annotations. Applications to the Framingham Heart Study data indicate that iFunMed is able to boost the detection of SNPs with mediation effects that can be attributed to regulatory mechanisms. Simulation experiments indicate that GRAD can improve the identification of genetic variants by increasing the average area under the precision-recall curve by up to 60\%. Real data applications to the Framingham Heart Study show that GRAD can select relevant genetic variants while detecting several transcription factors involved in specific phenotypical changes.