The variants that were significant (P < 5 × 10-8) in the meta-analysis, described above, were replicated in an independent sample of 1,306,354 unrelated individuals (414,055 cases and 892,299 controls) from 23andMe, Inc. Individuals within this replication sample were unrelated to individuals within 23andMe_307k. Detailed information regarding this replication sample is provided in the Supplementary Note. In summary, imputed genetic data was obtained for an unrelated European sample and the variants identified as genome-wide significant (P < 5 × 10-8) in the meta-analysis were analysed using logistic regression. The phenotype used in the 23andMe replication sample was the same used within the 23andMe_307k cohort, described above, with the responses to web-based surveys used to classify individuals that self-reported as having received a clinical diagnosis or treatment for depression classified as cases. We used Metal 56 to conduct an inverse variance-weighted meta-analysis of the significant variants (P < 5 × 10-8) identified in the previous paragraph (using UK Biobank, 23andMe_307 and PGC_139k) and the 23andMe replication cohort.
