was rs98545 (P=0.0019, OR=0.83, CI 0.75–0.94, MAF=0.15, intron 29). rs98545 lies 387 kb from rs1006737; they are in near linkage equilibrium (r2=0.06) and in our data their effects are additive in a logistic regression. Neither rs1006737 nor rs98545 were genotyped in the UK MDD or GAIN–MDD studies; imputation of GAIN–MDD genotypes showed weak association with rs1006737 (P=0.070, OR=1.10, CI=0.99–1.21, MAF=0.32) but not with rs98545 (P=0.52). We cannot exclude that the association of rs98545 has occurred by chance in our sample. Regional association plots for ADCY3, GAL and CACNA1C can be found in Supplementary File 2.
