Bayesian model averaging for IVW and MR-Egger

Uses Bayesian Model averaging to infer the posterior probability that the inverse variance weighted (IVW) or MR-Egger regression models (with or without heterogeneity) is the true data generating model. 

This method, developed for two-sample MR settings, requires the specification of priors for the regression model parameters and/or the probability that each model is "true". 

References

• Thompson JR, Minelli C, Bowden J, Del Greco FM, Gill D, Jones EM, et al. Mendelian randomization incorporating uncertainty about pleiotropy. Stat Med 2017; 36: 4627-4645.

Other terms in 'Model selection and averaging approaches for two-sample MR':

• Mixture of Experts (MR-MoE)
• Rucker framework
• Steiger filtering