Genetic architecture of alcohol use disorder using cross-trait genetic correlations and public next-generation sequencing studies

B. Todd Webb, Roseann E. Peterson, Robert M. Kirkpatrick, Kenneth S. Kendler, Michael C. Neale and Brien Riley

Recent advances in the study of human complex traits include 1) a growing catalog of robustly associated common variants found using genome-wide association studies (GWAS), 2) the ability to assess the contribution of rare variation through genome sequencing studies of both whole exomes (WES) and genomes (WGS), and 3) wide-spread cross-trait genetic correlation. For alcohol use disorder (AUD) and related phenotypes, there is an opportunity to discover novel rare variation and understand the overall pattern of evidence across the allele frequency spectrum including common and rare variation. One lesson from GWAS is that most robust associations do not create protein coding changes and some robust genome-wide significant variation does not reside in genes at all. As WES expands to include untranslated regions (UTRs) and WGS becomes affordable, there is a significant challenge in analyzing the variation that does not impact protein coding directly. Fortunately, there is a growing catalog of functional elements across the genome that can be interrogated to determine which harbor variants influencing AUD. For sequencing based analyses of rare variation, these functional elements showing enrichment can be leveraged to improve detection. As more variants associated with AUD are discovered and the catalog of traits with genetic correlations grows, determining how specific these variants are to AUD becomes more important. For example, do discovered loci influence a) addiction in general, b) alcohol consumption, or c) AUD specifically or do they non-specifically influence many physical or mental health outcomes. In addition to determining if loci, in aggregate or individually, can be considered primary or secondary AUD risk loci, understanding the direction of causation across traits and time is critically important to identifying opportunities for intervention. One powerful approach to investigate causative relationships is through Mendelian Randomization analysis. The overall goal of this project is to leverage existing twin, family, epidemiological, and molecular data in novel ways to a) discover loci harboring rare variants that influencing AUD without new molecular data generation, b) differentiate loci that influence common versus specific AUD liability factors, and c) understand the architecture of AUD using an integrated and iterative approach.