3.3 Sensitivity analysis and pleiotropy detection
We conducted a series of measures to avoid bias effects, including the sensitivity of MR analysis and the potential pleiotropy of IVs for each phenotype. The leave-one-out method indicated no outliers in Figure 4 (G), however, abnormal values were found in the remainder of Figure 4. However, these outliers affected the results could be ignored considering the other statistical methods (both the MR-Egger intercept test and MR-PRESSO global test suggested no statistical difference). Cochrane’s Q test indicated that no significant heterogeneity was observed among the selected SNPs except Genus. Oscillibacter. Because the IVW method was used as the main analysis method in this study, therefore, our results remain plausible. The detailed results are summarized in Table 2.