Home » AGAPS » AGAPS Tutorials » Mediation Tutorial

Mediation Tutorial

Output of mediation analysis is the file mediation.csv. The file containes 8 columns.

  1. var — SNP name
  2. mediator — mediator name
  3. Four death hazard ratios of the effect of minor allele:
    1. effect of SNP in univariate analysis (i.e., SNP + covariates)
    2. effect of SNP in bivariate analysis (i.e., SNP+mediator+covariates)
    3. effect of SNP in univariate analysis when mediator=0
    4. effect of SNP in univariate analysis when mediator=1
  4. Four p-values exactly corresponding to five HR from column 3
  5. Effects of the mediator on death:
    1. HR of mediator in bivariate model
    2. HR of mediator in univariate model
    3. Two respective p-values
    4. Two respective p-values
  6. OR and p-value of the effect of SNP on mediator (the model is Mediator=SNP+covariates)
  7. Chi2 of the hypothesis that the effect of SNP on death does not change after adding the mediator, higher |chi2| means that change is more significant:
    |chi2|=1 corresponds to (p=0.3) and |chi2|=2 corresponds to (p=0.05)
  8. Absolute value of chi2 from column 7
  9. A key showing whether all tests passed. passed = 1 or 2 if all tests below are passed. The tests are
    1. HR 1 from column 3 is significant
    2. HR 5 from column 6 is significant (passed=2 if p-value<0.05 and =1 if 0.05<p-value<0.1)
    3. Absolute value of effect size of SNP (i.e., |log HR|) of model 1 is larger than of model 2.
    4. |Chi2| from column 7 is larger than 1.