Sponsored by:
Since it opened access in 2012, UK Biobank (UKB) has supported researchers working in over 1375 institutes in 68 countries with various measurements and genotype information it collected from 500,000 research participants. However, about 9 million people were invited in order to obtain these 500,000 samples, which leads to a surprisingly low response rate of 5.5%. Previous investigation done by Fry et al. pointed out that UKB samples were healthier than the general population and were more likely to live in less socioeconomically deprived regions. We want to find out if there is genetic contribution to this selection bias. Because UKB refuses to construct sample weights or release relevant information, we built new sample weights to measure the likelihood of participation in the UKB, using demographic and geographic characteristics of respondents. Standard GWAS and auxiliary genetic analyses were then conducted with sample weight as the phenotype of interest. Our preliminary findings suggest hundreds of genetic loci are associated with survey participation and a SNP-heritability estimate of nearly 8%. We also calculated genetic correlations with many important phenotypes in the literature, including a 0.26 rg with educational attainment. In ongoing work, we re-estimate GWAS on educational attainment with control for survey participation weight in order to assess the robustness of earlier results.