Fig. 1: A schematic illustration for study designs and analyses.

We made large- and small-scale designs, each having four analyses with unrelated, first-, second- and third-degree relationships between discovery and target samples. Initially, we identified 288,837 unrelated individuals for whom any pairwise relationship was less than 0.05 (green). We also identified first- (dark brown), second- (brown) and third-degree (light brown) relatives using the information of kinship coefficients from the full UK Biobank sample. The analysis with the unrelated sample in the large-scale design was carried out for all unrelated individuals with available phenotypic information who were randomly divided into discovery (80%) and target data sets (20%). For the analysis with first-, second- or third-degree relatives in the large-scale design, the set of first-, second- or third-degree relatives were substituted with a set of the same number of randomly selected individuals in the analysis with unrelated sample. For the analysis with unrelated sample in the small-scale design, we used 6000 individuals (5000 as discovery, and 1000 as target sample), randomly selected from the analysis with unrelated sample in the large-scale design. However, with the analysis with first-, second- or third-degree relatives in the small-scale design, we selected 6000 individuals from the set of first-, second- or third-degree relatives, maximising the relationships among the selected individuals (see ‘Methods’).