This may reflect a complex genetic architecture in bipolar disorder, wherein one set of variants may be associated more with manic symptoms and another set with depressive symptoms. Variants associated more with mania (or psychosis) may have higher effect sizes, detectable at current bipolar disorder GWAS sample sizes, and may not be strongly associated with major depressive disorder. This could contribute to the observed higher heritability of bipolar disorder compared to major depressive disorder, and agrees with reports that most of the genetic variance for mania is not shared with depression (13, 14). In this case, meta-analysis of bipolar disorder and major depressive disorder cohorts would support variants associated more with depression, but not those associated more with mania. This is consistent with our findings, and with depressive symptoms being both the unifying feature of the mood disorders and the core feature of major depressive disorder.
