Cross-platform transfer of blood gene-expression models in Parkinson's disease — English provenance correction

CORRECTION NOTICE (published after this version): https://doi.org/10.5281/zenodo.23411084 supplies the corrected English provenance text and complete ZIP. The original Persian v3.1 preregistration and published v3.1.1 amendment govern if this translation differs. In this version, references claiming an English plan or step-zero file was saved/hashed before synthetic execution are incorrect: the original Persian files were saved/hashed before execution; the English files were created later. This record and its files retain their original timestamp. Complete English reference edition (v3.1.1-en) of the registered protocol v3.1 and implementation amendment v3.1.1. This is a faithful translation and presentation edition, with no methodological change. Original records: https://doi.org/10.5281/zenodo.23405371 and https://doi.org/10.5281/zenodo.23405802. The concept DOI remains 10.5281/zenodo.23405370. Author: Mohammad Javad Asgari, ORCID 0009-0002-2313-571X. Timing disclosure: one real PROBE/GSE57475 outer fold with all five models had already been completed and its results viewed before this English translation was prepared and published. This edition is not claimed to precede all real-data analysis, and the original registration timestamps are not backdated. No additional real-data training or scoring was performed for translation. No PPMI disease, sex, age or cell-marker score has been calculated. The package contains the complete English protocol, full preregistration text, numerical step-zero plan, solver amendment, synthetic plan and numerical results, historical warning-hold report, implementation amendment, and English GEO pilot report. All original code, config keys/values, seeds, thresholds, lambda/rho grids and JSON/JSONL evidence are preserved byte-for-byte. Machine checks compare the numeric inventories of all ten translated documents, verify technical identifiers and exact config/code preservation, and scan every file inside the ZIP for Arabic script and filenames marked _FA; all checks passed. Numeric matching supports but does not prove semantic equivalence; the text was also reviewed against the sources. Use the English texts to read the registered rules. The original Persian preregistration and the published implementation amendment take precedence if wording differs; no model, population, endpoint, gate or inference rule has changed. A source-to-translation hash map and an editorial-differences table are included. Historical source filename strings and freeze-time status flags remain in unchanged configuration/provenance files and are explained in the README. The final-model/preprocessing/population second public lock and the existing transfer gates remain required before PPMI disease scoring. No individual PPMI data, counts, private crosswalk or credentials are included. Start with README_ENGLISH_REFERENCE.md and PROTOCOL_v3_1_EN.md.