In examination of the follow-up samples by diagnosis, the novel variant, rs4583255[T], showed significant association with both schizophrenia and bipolar disorder (P = 0.0011 and 0.00026), with OR of 1.06 and 1.08, respectively (independent controls were used for the two analyses; see Supplementary Table 5). We also investigated association with bipolar disorder for variants that had shown genome-wide significant association with schizophrenia in our previous study14. Following correction for eight tests, rs12807809[T], near NRGN, was significantly associated with bipolar disorder (P = 0.0023) with an OR identical to that of the schizophrenia follow-up samples (OR = 1.09). The remaining schizophrenia susceptibility variants did not show nominally-significant association with bipolar disorder—yet OR confidence intervals for the two disorders overlapped for at least some variants at all loci (Supplementary Table 5).
