Data and code for: Lineage-specific regulatory deployment of conserved aging proteins across animal phyla

<p><strong>Self-contained data and code deposit</strong> for the manuscript:</p> <p>Tanigawa M (2026) "Lineage-specific regulatory deployment of conserved aging proteins across animal phyla." Submitted to <em>Nature Aging</em>.</p> <p><strong>Summary.</strong> This deposit contains all input data, analysis scripts, intermediate outputs, and supplementary tables required to reproduce every numerical result reported in the manuscript. The study computes Kendall's W concordance for protein identity and four regulatory indicators (gene-level expression, isoform usage, splice-site Evo 2 7B embeddings, intron length) on 87 aging and comparator genes across five animal phyla (Nematoda, Insecta, Mammalia, Cnidaria including <em>Hydra vulgaris</em>, and Echinodermata). The central finding is that the magnitude of the protein–regulation decoupling is not aging-specific (permutation p = 0.53), but the cross-phylum allocation of regulatory constraint among aging subcategories is aging-specific (Friedman p = 0.007), with nematodes investing in DNA-damage response and mammals in mTOR nutrient-sensing.</p> <p><strong>Contents.</strong></p><ul><li>52,280 splice-site sequences from 87 genes across 140 species and corresponding Evo 2 7B embedding HDF5 files (~1.4 GB)</li><li>Per-gene Mantel results, protein identity (perc_id) tables, and dN/dS results</li><li>Salmon transcript quantification (quant.sf) for 6 RNA-seq samples (<em>Drosophila</em> modENCODE, <em>C. elegans</em> N2 adult)</li><li>Isoform diversity metrics (Shannon entropy and dominant-isoform fraction)</li><li>Gene-level expression and intron-length tables (multi-indicator extension)</li><li>Bayesian network analysis results (1,000 bootstrap iterations per phylum)</li><li>5-phylum extension data (Cnidaria and Echinodermata orthologs, phylogenetic distances, Evo 2 embeddings, Mantel results, leave-one-phylum-out outputs)</li><li>Per-gene canonical case study (<em>daf-2</em>/IGF1R, <em>daf-16</em>/FOXO, <em>let-363</em>/mTOR)</li><li>Hallmarks-of-aging mapping (López-Otín <em>et al.</em> 2023; Supplementary Table S19)</li><li>Friedman robustness panels for the subcategory investment profile (Supplementary Table S20: baseline, bootstrap CI, leave-one-subcategory-out, leave-one-phylum-out Spearman, within-subcategory permutation null)</li><li>39 Python analysis scripts</li></ul> <p><strong>Relationship to a separate study.</strong> The splice-site sequences and Evo 2 7B embeddings were originally generated for a separate methodological study (Tanigawa &amp; Iwaki, submitted to <em>Molecular Ecology</em>) that benchmarks five genomic foundation models on category-level splice-site evolutionary constraint. The two studies share only the splice-site dataset and have no overlap in analytical methods, statistical tests, or biological conclusions. To make the present deposit fully self-contained, copies of the shared data are included under <code>data/shared_splice_data/</code>. Reproducing every analysis requires only the files in this deposit; no other Zenodo record needs to be consulted.</p> <p><strong>License:</strong> CC-BY 4.0</p>