SERUMSPEC-PCa: analysis code, extended tables and additional figures for a tumour-type specificity audit of serum microRNA classifiers in prostate cancer (GEO GSE112264)

This deposit contains the complete analysis code, frozen result files, extended data tables and additional figures supporting a computational reanalysis of GEO series GSE112264 (2,565 mature microRNA probes; 1,591 sera: 809 prostate cancers, 282 controls, 500 patients with ten other cancer types) on the 3D-Gene GPL21263 platform. The study evaluates tumour-type specificity as an explicit endpoint for organ-specific circulating biomarker classifiers. It comprises leakage-controlled nested 5×5 cross-validation with per-fold decision thresholds, cross-application of the frozen case–control classifier to other-cancer sera, deliberate-leakage optimism quantification, conformal coverage checks, sample-quality control including a haemolysis index, independent renormalisation, moderated differential expression with false discovery rate control, covariate-adjusted analyses using clinical metadata recovered from GSE211692, microRNA target and pathway enrichment, protein–protein interaction networks, unsupervised structure analysis and feature-selection stability. The deposit also includes a structured comparison of nine published serum and plasma microRNA diagnostic studies in prostate cancer, giving the performance each reported and whether it was evaluated against other cancer types. All numerical values are traceable to the result files included here. Random seeds are fixed, the expression matrix is rebuilt from the public GEO source files with SHA-256 checksum verification, and the software manifest declares the three execution environments used. Associated article: Cancer Genomics & Proteomics (submitted).