Beyond The Genome: Integrative Multi-omic Analysis Of Complex Traits Using Biobank Data

Human complex traits arise from the intricate interplay between genetic, molecu- lar, and environmental influences. This thesis leverages large-scale biobank data to move beyond the genome, integrating genomic, transcriptomic, proteomic, and phenotypic layers to dissect the biological mechanisms underlying complex traits and to enhance biomarker discovery. In the first study, we investigated the molecular basis of the genetic correlation between body mass index (BMI) and brain morphology using the UK Biobank imaging genetics resource. By combining genome-wide association data with brain expression quantitative trait loci (eQTLs), we identified 21 genes whose genetically regulated expression in the brain pleiotropically influences both BMI and regional brain structure. Fine-mapping, colocalization, and epigenetic annotation high- light causal variants in loci such as TUFM and VPS11, implicating mitochondrial translation and microglial regulation in the neurobiological architecture of obesity. In the second study, we developed a framework to remove inherited genetic effects from plasma protein levels—genetic adjustment—to refine biomarker prediction. Using proteomic and genomic data from nearly 40,000 UK Biobank participants, we showed that genetically adjusted proteins display stronger associations with 37 diseases and with lifestyle or environmental exposures, corresponding to a 30% median gain in statistical power for biomarker discovery. Multi-protein models derived from adjusted proteins outperform standard models across multiple con- ditions, illustrating how subtracting genetic variability enhances detection of bio- logically relevant, non-genetic signals. Together, these studies demonstrate that biobank-scale multi-omic integration can reveal the molecular mechanisms linking genotype, intermediate molecular pheno- types, and disease. They advance a generalizable paradigm for moving beyond the genome toward a more complete, mechanistic, and predictive understanding of human health and disease.