therefore not analyzed. SNPs rs10994336 and rs10994397 in the 5′ region that had previously associated with bipolar disorder but not ankyrin G expression were tested in the second stage (Ferreira et al., 2008; Schulze et al., 2009; Scott et al., 2009; Sklar et al., 2011; Tesli et al., 2011). Although we started with the prior hypothesis that the association between rs9804190 would be in the same direction as the observed association with bipolar disorder (the major C allele conferring risk), we report the two-sided association test results for rs9804190 so that the p-values are comparable to those reported for other ANK3 SNPs for which we did not have a strong prior hypothesis.
