Analysis code for: Measured childhood BMI and type 2 diabetes: the direct effect reverses on conditioning for adult BMI in a triangulated multivariable Mendelian randomisation study

Le résumé fourni par la source

Analysis code for the Study: "Measured childhood and adult adiposity and cardiometabolic disease: a triangulated multivariable Mendelian randomisation study". This archive contains every R script underlying the analyses reported in the manuscript and its supplementary material: instrument selection, univariable and multivariable two-sample Mendelian randomisation, bidirectional MR, Bayesian colocalisation (coloc.abf and coloc.susie), a confounder-pleiotropy screen, and the sensitivity analyses (Richardson lifecourse instruments, UK-Biobank-free adult BMI, DIAMANTE 2022 outcome, weak-instrument-robust estimators, MRBEE, parametric weak-instrument simulation, leave-one-out, minimum detectable effects), together with the scripts that draw every figure and assemble the supplementary tables. Summary statistics are not redistributed; the README lists their public sources and expected locations. On 4 September 2026 the complete pipeline was re-run from the raw summary statistics and reproduced every reported value; the README documents this reproducibility record and the two known differences (server-side LD reference-panel drift in coloc.susie, at most 0.0011 in posterior probability, and a corrected UK Biobank denominator in the sample-overlap table).

Ce résumé expose les affirmations des auteurs. BNTIC ne l’interprète pas comme une validation indépendante des résultats.