MRforAD

Package overview

The MRforAD is an R package designed for Mendelian Randomization (MR) analysis, focusing on Alzheimer's Disease (AD). Its main purpose is to be used to study causal relationships in AD using genetics data. It uses publicly available GWAS summary statistics, such as those from Jansen et al. (2019), and supports integration with user-provided ADNI data from those with approved access.

Introduction

Mendelian Randomization (MR) is a statistical method in genetic epidemiology used to infer causal relationships between casual relationships between an exposure and an outcome by using genetic variants as instrumental variables. This approach is particularly interesting for studying complex disease like Alzheimer's disease (AD), where observational studies may be confounded by environmental factors.

The Alzheimer's Disease Neuroimaging Initiative (ADNI) provides a rich dataset including genetic, clinical, and imaging data; making it a primary candidate for MR analyses. However, ADNI data access is restricted, requiring approval through the LONI Image and Data Archive. With that being said, the development of an R package focuses on using publicly available summary statistics for AD, while providing functionality for any user with ADNI access.

Bioinformatic analyses background

MR relies on three key assumptions: the genetic instrument is associated with the exposure (relevance), is independent of confounders (independence), and affects the outcome only through the exposure (exclusion restriction). For AD, this method can help to identify causal risk factors, such as genetic variants influencing AD risk through pathways like immune response or lipid metabolism.

The ADNI cohort, with is longitudinal data on cognitively normal, mild cognitive impairment and AD/dementia participants, is ideal for such analyses. But its restricted access poses a challenge. To address this, the package uses publicly available AD GWAS summary statistics, ensuring broad accessibility while maintaining relevance for ADNI users.

Data and functionality

Since ADNI data requires approved access through the LONI Image and Data Archive (http://adni.loni.usc.edu), this tool cannot include it directly. However, i used directly summary statistics from Jansen et al. (2019), available at https://ctg.cncr.nl/software/summary_statistics. I included functionns to download these datasets, perform MR analyses (including instrument selection, harmonization, and sensitivity checks), and visualize results by making is suitably for the users to present the results.

Core functions:

• download_gwas_data_url(url): this f(x) downloads GWAS summary statistics from specified URLs.
• get_ad_gwas_summary(study_id): this fx() retrieves data from the GWAS/NIAGADS API, which provides programmatic access to publicly AD GWAS summary statistic.
• run_mr_pipeline(exposure_data, outcome_data): this f(x) performs the full MR pipeline, including instrument selection, harmonization, MR analysis (e.g., IVW, MR-Egger, weighted median), sensitivity analyses (e.g., MR-egger intercept, heterogeneity tests), and visualization.
• plot_mr_results(mr_results): this f(x) generates standard MR plots like scatter, forest and funnel plots.
• prepare_mr_data(exposure_file, outcome_file): this f(x) loads and prepares user-provided data for analysis.