DESCRIPTION
Description
R package for two-sample Mendelian randomization analysis using mixture models. For numerical stability, we recommend using summary statistics in the standardized scale, i.e.
- 1) for continuous phenotype, the beta and SE when both the genotypes and phenotypes are standardized to have variance 1;
- 2) for binary phenotype, the beta and SE when the genotypes are standardized to have variance 1.
Causal estimates are interpreted as standard deviation (SD) unit increase (continuous outcome) or log-OR (binary outcome) of Y per SD unit increase in X (assumed to be continous). If standardized-scale summary statistics are not available, users may use z-statistics and effective sample size to transform their summary statistics to the standardized scale. See Example or type ?MRMix for more information.
System requirements
MRMix can be used on any operating system. R needs to be installed. Package devtools is required for the installation.
Installation
devtools::install_github("gqi/MRMix")