Reproducible code for signal-specific fine-mapping of the APH1B locus in Alzheimer disease

Abstract

This software archive contains analysis and figure-generation code for signal-specific fine-mapping of the APH1B locus in Alzheimer disease. Version 2026.10.10 synchronizes the publication display with the revised manuscript: Figure 1 names the specified multiple-testing correction, Figure 3 identifies its highest-PIP variant subset and reports expression effects in a source-specific table, Figure 4 distinguishes positive-definite LD settings with filled markers, and the FinnGen display explicitly identifies the whole-blood eQTL comparator. This FinnGen analysis is not a matched replication test of the cortical two-signal model. The graphical abstract retains the author-approved vertical three-column manual vector design, brain illustrations, arrows and colours, while focusing its text on the two resolved cortical expression signals. The rs117618017-centered signal shares the disease association; the rs12708477-centered signal is distinct. Walker supports molecular sharing without unique variant resolution and reuses the disease GWAS. FinnGen supplies marginal clinical association, and neuropathology associations are corrected across eight correlated endpoints. Expression, splicing and Thr27Ile coding effects remain to be distinguished. Canvas: 254 x 317.5 mm; Arial and Arial Narrow. No AI image generation was used. Publication-display reproduction completed with exit code 0. Scientific source statistics, numerical analyses, correction families and analysis scripts are unchanged. Historical execution evidence and the targeted Walker credible-set verification are preserved; the historical full-analysis terminal process exit code remains unrecorded. The canonical portable display script is byte-identical to the submitted Figure_reproduction_code.txt. This code-only archive includes code, environments, input specifications and verification documentation. It contains no manuscript, rendered figures, result tables, external or participant-level data, prepared inputs, GWAS/QTL summary statistics, LD matrices or font files. Scientific analysis reproduction requires the documented external/prepared inputs; current figure reproduction requires the supplementary workbook supplied separately. Historical plotting scripts are retained as provenance; the current publication route uses portable/reproduce_publication_figures.py. Provider access terms apply. Statistical sharing does not establish molecular mediation.