2.5 Statistical analysis
Bonferroni-corrected test was used to establish the significance thresholds for the primary MR results, which considered the number of bacteria in each feature level (phylum: 5.56×10-3, class: 3.13×10-3, order: 2.5×10-3, family: 1.56×10-3 and genus 4.2×10-4, respectively). The P value of MR results less than the Bonferroni-corrected threshold was regarded as statistical significance. All statistical analyses were conducted by the Two-Sample MR package (version 0.5.7) of R software (version 4.3.1). In addition, we used an online power calculation tool for MR to evaluate the statistical power of causal effect estimates [18], which could be accessed by the website:https://shiny.cnsgenomics.com/mRnd/. Two-sided P < 0.05 was statistically significant in MR results.