Reproducibility archive: GPC2 expression and survival in pediatric neuroblastoma, and an audit of assay reuse between public cohorts

Version 2.0.0 supports two articles: "GPC2 expression and adjusted survival associations in pediatric neuroblastoma using public data" and "Undocumented assay reuse between two public neuroblastoma expression cohorts". The version 1.0.0 archive (original TARGET/GTEx expression and SEQC survival analysis) is kept unchanged in this record as "Data S1 GPC2 reproducibility archive.zip"; the new material is in "GPC2 reproducibility archive v2.zip". GPC2 survival analyses (first article). The archive contains the survival data and code for the post hoc pediatric restriction of SEQC (A1) and the amended exploratory analysis of the NRC cohort (GSE85047, A2), an independent implementation of the stratified Efron partial likelihood with a GPC2 × ln(t/5 years) term, the recomputed coefficients and covariance matrices, GDC age verification for 153 TARGET patients, and the figure scripts. No GPC2-outcome model was fitted in the Kocak series because most of its assays are shared with SEQC. Assay-identity audit (second article). The array barcode was extracted from the raw-file header (Agilent Feature Extraction FEPARAMS block) of all 498 SEQC microarrays in GEO GSE49710 and matched with the barcodes of the 649 samples of the Kocak series (GSE45547). In total, 468 SEQC arrays matched 468 distinct Kocak assays. Three complete raw-file pairs were byte-identical after decompression, and matched pairs agreed on age group (468/468), MYCN status (463/463 with data in both series) and INSS stage (467/468). The archive contains the header-only extracts of the 498 SEQC raw files, the three complete raw-file pairs, the barcode and clinical correspondence tables, the Kocak partition by shared and unshared assays, and a script that recomputes every audit number. All inputs are public, deidentified data from NCBI GEO, UCSC Xena Toil, the NCI Genomic Data Commons and public R2 clinical exports. No person was contacted or identified. A non-matching barcode does not establish a different participant. Users should cite the original data sources. The license covers the compilation, code and derived outputs. File note: the version 2.0.0 archive was replaced on 30 October 2026 to add the pediatric SEQC analysis script and its results and to update the title in the citation file; no previously reported result changed.