run_mr_analysis: Run MR Analysis for Multiple Outcomes

In autoMR: Automated Mendelian Randomization Workflows and Visualizations

Run MR Analysis for Multiple Outcomes

Description

Performs causal inference analysis using multiple Mendelian Randomization (MR) methods across one or more outcomes and exposures. Returns a combined results data frame. To save the output, use standard R functions such as write.csv() or saveRDS() on the returned object.

Usage

run_mr_analysis(
  MR_input_data,
  outcome.form = NULL,
  use_ivw = TRUE,
  use_raps = TRUE,
  use_median = TRUE,
  use_egger = TRUE,
  use_mr_presso = TRUE,
  use_mr_horse = TRUE,
  use_mr_grip = TRUE,
  NbDistribution = 1000,
  SignifThreshold = 0.05,
  mr_horse_n_iter = 5000,
  mr_horse_n_burnin = 1000,
  mr_grip_parameters = NULL
)

Arguments

MR_input_data	Harmonised MR input data frame. Must contain Outcome and Exposure columns.
outcome.form	Character vector indicating the effect scale for each outcome: "Beta", "OR" (odds ratio), or "HR" (hazard ratio). A single value is recycled across all outcomes. Defaults to "Beta".
use_ivw	Logical; whether to run the Inverse Variance Weighted (IVW) method. Default is TRUE.
use_raps	Logical; whether to run the Robust Adjusted Profile Score (MR-RAPS) method. Default is TRUE.
use_median	Logical; whether to run the Weighted Median method. Default is TRUE.
use_egger	Logical; whether to run MR-Egger regression. Default is TRUE.
use_mr_presso	Logical; whether to run the Mendelian Randomization Pleiotropy RESidual Sum and Outlier (MR-PRESSO) method. Default is TRUE.
use_mr_horse	Logical; whether to run the MR-Horse method. Default is TRUE.
use_mr_grip	Logical; whether to run the Generalized Regression with Instrument Pairs (MR-GRIP) method. Default is TRUE.
NbDistribution	Integer; number of simulated distributions for MR-PRESSO. Default is 1000.
SignifThreshold	Numeric; significance threshold for the MR-PRESSO outlier test. Default is 0.05.
mr_horse_n_iter	Integer; number of Markov chain Monte Carlo (MCMC) iterations for MR-Horse. Default is 5000.
mr_horse_n_burnin	Integer; number of MCMC burn-in samples for MR-Horse. Default is 1000.
mr_grip_parameters	Accepted for API compatibility; not used by MR-GRIP.

Value

A data frame combining results across all outcomes and exposures. Each row represents one outcome-exposure pair. Columns include estimates, confidence intervals (CI), and p-values for each method, together with diagnostic flags (e.g., F-statistic below 10, significant heterogeneity). Use write.csv() or saveRDS() to save the returned object.

Examples

data("fi_49item")
input1 <- harmonize_mr_data(df = fi_49item)$input_df
outcome1 <- run_mr_analysis(
  MR_input_data     = input1,
  outcome.form      = "Beta",
  use_ivw           = TRUE,
  use_raps          = FALSE,
  use_median        = FALSE,
  use_egger         = FALSE,
  use_mr_presso     = FALSE,
  use_mr_horse      = FALSE,
  use_mr_grip       = FALSE,
  NbDistribution    = 1000,
  SignifThreshold   = 0.05,
  mr_horse_n_iter   = 5000,
  mr_horse_n_burnin = 1000,
  mr_grip_parameters = NULL
)


data("fried_frailty")
input2 <- harmonize_mr_data(df = fried_frailty)$input_df
outcome2 <- run_mr_analysis(
  MR_input_data     = input2,
  outcome.form      = "OR",
  use_ivw           = TRUE,
  use_raps          = TRUE,
  use_median        = TRUE,
  use_egger         = TRUE,
  use_mr_presso     = TRUE,
  use_mr_horse      = TRUE,
  use_mr_grip       = TRUE,
  NbDistribution    = 1000,
  SignifThreshold   = 0.05,
  mr_horse_n_iter   = 5000,
  mr_horse_n_burnin = 1000,
  mr_grip_parameters = NULL
)



data("merged_data")
input3 <- harmonize_mr_data(df = merged_data)$input_df
outcome3 <- run_mr_analysis(
  MR_input_data     = input3,
  outcome.form      = c("Beta","OR"), ## First outcome use Beta and second outcome use OR
  use_ivw           = TRUE,
  use_raps          = TRUE,
  use_median        = TRUE,
  use_egger         = TRUE,
  use_mr_presso     = TRUE,
  use_mr_horse      = TRUE,
  use_mr_grip       = TRUE,
  NbDistribution    = 1000,
  SignifThreshold   = 0.05,
  mr_horse_n_iter   = 5000,
  mr_horse_n_burnin = 1000,
  mr_grip_parameters = NULL
)

autoMR documentation built on June 12, 2026, 9:07 a.m.