LDSC regression 10 was used to calculate pairwise genetic correlations (rG) to determine the extent of overlap in the genetic architectures across the three non-overlapping cohorts that contributed to our meta-analysis, i.e. between the UK Biobank, PGC_139k and 23andMe_307k depression analyses. There was a strong genetic correlation between each of these cohorts. The rG between PGC_139k and 23andMe_307k was 0.85 (s.e = 0.03). The UK Biobank had a rG of 0.87 (s.e = 0.04) with PGC_139k. A similar rG was also found between UK Biobank and 23andMe_307k (0.85, s.e = 0.03). This was despite UK Biobank using a broader phenotype based on self-reported help-seeking behaviour compared to the self-declared clinical depression phenotype of 23andMe_307k and the primarily clinically obtained MDD phenotype of PGC_139k.
