A putative causal effect of depression on neuroticism was detected using the IVW regression analysis MR test at the 1% significance threshold after false discovery rate (FDR) correction (beta = 0.146, s.e. = 0.039; PFDR = 2.29 × 10-3; Supplementary Figure 4). However, there was also evidence of variant heterogeneity (P = 6.01 × 10-3), due to global horizontal pleiotropy, requiring additional sensitivity tests to examine the consistency of the effect. The additional sensitivity tests both had effects in the same direction as the IVW test (weighted median beta = 0.119, s.e. = 0.047; MR Egger beta = 0.050, s.e. = 0.235). There was also evidence of a putative causal effect of depression on ever vs. never smoked (beta = 0.285, s.e. = 0.077; PFDR = 2.29 × 10-3; Supplementary Figure 5), with no evidence of variant heterogeneity (P = 0.14). Both the putative causal effects of depression on neuroticism and depression on ever vs. never smoked remained consistent (P < 6.46 × 10-4) in the ‘leave one variant out’ IVW analysis indicating that the observed effect was not driven by a single outlying variant.
