Construction-Matched Controls Reveal Aneuploidy Confounding in Gene-Set Copy-Number Biomarkers Across 79,622 Patients: Reproducibility Package

Version 2.0 reproducibility package for a construction-matched evaluation of copy-number gene-set scores. It contains complete deposited inputs; 10,000-draw cohort- and gene-set-specific matched nulls hard-matched on size, platform coverage, pre-encoding missing-call count, chromosome-specific 25-Mb locus and deletion frequency; finite-sample empirical and multiple-comparison inference; crossed hierarchical GDSC/PARP models; clinically augmented prognostic models with bootstrap validation and calibration; row-wise source provenance; analysis outputs; and a one-command verification/full-reanalysis entry point. The 17 pathway/tumor-suppressor comparators are study-curated core subsets anchored to explicit KEGG, Reactome or COSMIC identifiers, not complete database exports. PROfound and FoundationOne-derived entries are analytical gene sets only and do not recreate or validate the original trial biomarker or complete companion diagnostic.