mr-metabolite-discordance: analytical choices that drive discordant two-sample Mendelian randomization results (a specification-curve analysis of three metabolite-disease pairs)
Year: 2026
Type: software
Abstract: Version 1.2.0. Cross-references in MANIFEST.md now follow the numbering of the current manuscript: main-text figures are cited as Figures 1-7, and material that is not a main-text table is named by description and marked (Supplementary) rather than by a table number. The dataset-identifier pointer in README.md now cites the technical-parameterisation table (Supplementary) and Additional file 1. Added NOTICE_simulation_labels.md, recording that the fields ivw_fpr and raps_fpr report empirical power at theta = 0.10, not false-positive rates. Title aligned with the manuscript. Analysis pipeline and derived results for a study of why independent two-sample Mendelian randomisation (MR) studies of the same metabolite-disease pair reach opposite conclusions. Three contested pairs (glycine-coronary artery disease, acetate-heart failure, omega-3/DHA-atrial fibrillation) are re-analysed with ten analytical choices varied one at a time on fixed data. The repository contains the full pipeline, including pure-Python re-implementations of MR-PRESSO and MR-RAPS (validated by simulation and benchmarked against the canonical R packages), MR-Egger with an explicit orientation (flip) step, four FDR correction families, summary-level colocalisation, and the derived result files backing the tables and figures of the associated manuscript. No R was used. All input data are publicly available summary statistics retrieved from OpenGWAS and the GWAS Catalog; no individual-level or identifiable data are included.
Author: Lei, Guangtao
Institution: Second Affiliated Hospital of Nanchang University
Language: English
Subfield: Genetics
Domain: Life Sciences
Open Access status: green
OpenAlex ID: W7214701696
