Extended Data Fig. 3: MR effect on atrial fibrillation against the MR effect on HFpEF and HFrEF for genes that pass the Bonferroni threshold and including HFpEF suggestive genes.

The plot is shown for HFpEF findings that pass the Bonferroni-adjusted threshold of p-value \(< 2.06\times {10}^{-6}\) (in red) and for findings that pass the false discovery rate (FDR) of 5% (p-value \(< 6.80\times {10}^{-4}\)) (in blue). The sample size used to derive statistics for the MR effects on HFpEF is 27,579 participants with HFpEF, 27,799 participants with HFrEF, and 367,267 control individuals. The sample size used to derive statistics for the MR effects on atrial fibrillation is 91,703 cases and 776,176 controls. The predicted model is y = 0.0079 + 0.83x. The correlation between the MR betas for atrial fibrillation and HFpEF is 0.93. The errors bars represent the 95% confidence interval, defined as 1.96*standard error. All tests were two-sided without adjustment for multiple testing.