Bycroft, C. et al. Nature 562, 203–209 (2018).

How are human genetic variations linked to disease phenotypes? In order for this question to be answered in any detail, large amounts of patient data are needed. The UK Biobank is building such a unique resource: researchers involved in the project collected genotype data on ~500,000 individuals in conjunction with phenotypic data such as measures of physical activity, blood, saliva and urine biomarkers, in addition to MRI images of heart and brain. Bycroft et al. analyzed the quality of the array-derived genotype data, as well as their population structure and relatedness. Imputation allowed them to increase the number of variants to 96 million. In particular, the imputation of variation at human leukocyte antigen (HLA) genes provided the researchers with signal for the association of particular HLA gene alleles and several diseases.