CochleaConsensus v1.0: a uniformly reanalysed cross-study resource of animal hearing-loss models with a mouse–zebrafish regeneration contrast — data package

A systematic curation and uniform reanalysis of public GEO transcriptomic datasets for animal hearing-loss models: mouse as the main species, plus a zebrafish regeneration module. Unlike data portals that host and visualise individual datasets (e.g. gEAR: Orvis et al., Nature Methods 2021, PMID 34172972; Hertzano & Mahurkar, Human Genetics 2022, PMID 35235019), this resource provides (1) sample-level adjudication of every curated series, (2) uniform bulk reanalysis with per-insult cross-study consensus response genes, (3) cell-type-level pseudobulk differential expression from single-cell series, and (4) a formal mouse–zebrafish 1:1-ortholog concordance test for regeneration-related responses. The curated core comprises 88 mouse series (2,242 samples) plus a 10-series zebrafish regeneration module (176 samples; 98 series, 2,418 samples in total). Sample-level adjudication found that three genetic-layer series (GSE40207, GSE172286, GSE111668) contain no inner-ear samples; these are retained in the curation table, flagged as unusable, and excluded from all reanalysis. Headline numbers: full census 172 series; 86 independent studies; 110 usable contrasts, 77 bulk contrasts successfully reanalysed; cross-study consensus genes (strong+consensus, ≥2 studies): GENETIC 3,729; FISH-regen 3,687; REGEN 838; NOISE 444; OTOTOX 198; AGING 1 (Ch25h, descriptive only). Single-cell layer: 17 cell-level series totalling 624,714 cells, 58 cell-type DE tables. Ortholog concordance vs zebrafish regeneration (7,983 1:1 pairs): REGEN 0.563, GENETIC 0.546, NOISE 0.534, OTOTOX non-significant, AGING 0.475 (no concordance). Version v1.0, freeze date 2026-10-29. File inventory with MD5 checksums: see MANIFEST.csv in this record. Small result tables are packed into per-category ZIP archives whose contents reproduce the package file-for-file as listed in MANIFEST.csv; large objects (gene_profiles_long.csv, gene_profiles.json), the curation table, README and MANIFEST are individual files.