| dc.description.abstract | Decades of genome-wide association and linkage studies have resulted in tremendous progress in mapping genetic loci underlying disease risk, however, a gap between estimated heritability and mapped heritability, so called “missing heritability”, remains for most complex traits. Novel gene-mapping techniques may help uncover some of this missing heritability. I present two such approaches here, polygenic risk score adjusted linkage analysis and transcriptome prediction association analyses. First, in a family highly enriched for non-syndromic cleft lip and palate (NSCLP), I applied polygenic risk score adjusted linkage, utilizing variants from published genome-wide association studies (GWAS) of NSCLP to create a polygenic risk score for the trait, then generating individual-specific liability classes for each family member to account for their individual polygenic risk. Using this approach, I identified a rare missense variant in PDGFRA that results in a 7.4-fold increase in likelihood of displaying the trait. Second, I applied S-PrediXcan, a method for transcriptome prediction association analyses, to summary statistics from two GWAS of carotid intima media thickness (cIMT) and compared these results to a differential expression analysis for the trait, identifying TLN2 as a novel gene associated with cIMT and providing insight into utility of the two transcriptome-based approaches. Finally, I completed meta-GWAS and transcriptome prediction analyes of type 2 diabetes and lipid traits in Hispanic/Latino populations, identifying 12 novel single variant loci, 51 novel genes via transcriptome prediction, and proposing a mechanism of effect for 50 known loci across the traits. These studies demonstrate the potential of innovative approaches such as polygenic risk score adjusted linkage and transcriptome prediction to identify loci underlying the missing heritability of complex disease. | |
