We have three novel methods, one for detecting regions of the genome which influence a trait, one for creating risk prediction models, one for estimating the total contribution of rare genetic variants on traits. We wish to apply these to genetic data for traits such as height, BMI and Type 2 diabetes. We have demonstrated our methods on medium-sized data (5-15k individuals), however, Biobank access will allow us to apply to very-large data. We will first use the data to refine the methods, then perform the analysis proper; the result will be a list of genetic regions significantly associated with each trait, which will improve our understanding of their genetic architectures, a set of prediction models (i.e., SNP effect sizes), which allow us to better predict individual’s phenotypes from their genotypes, and estimates of rare SNP heritability (the total contribution to a trait of rare genetic variants) which provide a better understanding of genetic architecture. We make all methods freely-available to other researchers via LDAK (www.ldak.org).
