Stratified meta-analysis of rodent diabetic neuropathy nerve transcriptomes identifies a down-regulated structural core

Abstract

This record contains the derived data and analysis code for the study "Stratified meta-analysis of rodent diabetic neuropathy nerve transcriptomes identifies a down-regulated structural core" (manuscript submitted to Disease Models & Mechanisms, 2026). Twelve rodent (mouse and rat) microarray and RNA-seq datasets of diabetic peripheral neuropathy nerve tissue (211 samples after quality control; sciatic nerve and dorsal root ganglion) were re-processed from raw data and pooled per gene by REML random-effects meta-analysis, globally and in six tissue, stage and model strata. A cross-stratum direction vote defines a 484-gene candidate pool, a 225-gene direction-consistent core (79 up, 146 down) and a 57-gene robust core retained under all 11 single-study omissions. A design-matched simulation calibrates the REML z reference (type I error 8.8 % at nominal 5 % over 30 replicates), truncated Knapp–Hartung counts serve as the near-nominal sensitivity, and the core is projected onto one human sural-nerve RNA-seq dataset as an external directional reference (no detectable gene-level directional conservation). DPN_rodent_meta_v2.zip holds the per-study differential-expression and orthology tables, meta-analysis inputs and outputs, direction-vote and robustness results, enrichment and tissue-composition analyses, the calibration benchmark, all figure and supplementary-table source code, and per-file SHA-256 checksums (SHA256_MANIFEST.tsv, also uploaded alongside). README.md and DATA_AVAILABILITY.md describe the layout and the GEO accessions of the underlying raw data, which are not redistributed here. Data and documentation are released under CC BY 4.0, code under MIT. This version supersedes the original release and is the version analysed in the submitted manuscript.