To illustrate the general procedure, consider the single-trait input GWAS for neuroticism in ELSA and EGCUT. Supplementary Table 8 shows that the largest meta-analysis of neuroticism (NEURO1) yielded a final sample of N = 484,560 individuals by combining data from UKB (N = 361,688), 23andMe (N = 59,206) and a previously published study (N = 63,666). Since the column does not indicate any overlap with ELSA, the single-trait input GWAS for neuroticism in ELSA is the set of summary statistics from this meta-analysis. EGCUT, however, is listed in Supplementary Table 8 as overlapping with the NEURO1 meta-analysis. The reason is that EGCUT contributed to the summary statistics of the previously published study (it is one of the cohorts in de Moor et al.66). To eliminate overlap, EGCUT’s single-trait input is therefore generated by meta-analysing the summary statistics from UKB (N = 361,688) and 23andMe (N = 59,206) only. This restricted meta-analysis is listed in the table as NEURO2. Similarly, the largest single-trait input GWAS for neuroticism includes the UKB, so all three UKB partitions are listed as overlapping with it.
