mrcieu / mr-e-values

Code to calculate effect estimates and descriptive statistics to be used as inputs for the MR-E-Values shiny app (https://mark-gibson.shinyapps.io/mr-e-values_shiny_app/), using the applied example of the potential causal effect of smoking behaviour on schizophrenia.

The MR-E-Value: Smoking and Schizophrenia

This pipeline produces linear regression and Mendelian randomisation estimates for the potential causal relationship between smoking initiation and schizophrenia, using UK Biobank data, to be used to calculate the MR-E-Values.

Directory Structure

The project folder has the following structure:

MR-E-Value/code
MR-E-Value/code/a_extract_data
MR-E-Value/code/b_process
MR-E-Value/code/c_analysis
MR-E-Value/data
MR-E-Value/output

Input Files

data/data-mr.txt

This file contains data for schizophrenia (ICD 10 codes F20-F24,F28-29), smoking (ever versus never), risk-taking (risk-taker versus not) and beta-weighted genetic risk scores for smoking (Liu, 2019) and risk-taking (Linner, 2019). This file combines data created for Mendelian randomisations by two other pipelines. These are MRCIEU/MultiverseMR_Smoking_Schizophrenia and MRCIEU/Confounder-MR (risk-taking).

Scripts

a_extract_data

This section of the pipeline extracts dates for schizophrenia diagnoses and calculates whether these were before or after baseline for each individual.

b_process

This section of the pipeline excludes individuals with schizophrenia diagnoses before baseline and those with missing data for smoking initiation.

c_analysis

This section of the pipeline conducts a linear regressions of schizophrenia on smoking initiation and an analogous MR. These effect estimates can then be entered into my shiny application (https://mark-gibson.shinyapps.io/mr-e-values_shiny_app/) which conducts the MR-E-Value analysis. This section of the pipeline also conducts MRs of both smoking initiation and schizophrenia on risk-taking to investigate it's strength as a potential confounder.