Data and Code (version 3): "Cross-phylum multi-layer analysis reveals conserved aging proteins but divergent regulatory deployment"
收藏资源简介：
<p><strong>Manuscript citation.</strong> Tanigawa M (2026) Cross-phylum multi-layer analysis reveals conserved aging proteins but divergent regulatory deployment. <em>Submitted to PLOS Biology.</em></p> <p><strong>Concept DOI (all versions):</strong> 10.5281/zenodo.19364280<br> <strong>License:</strong> CC-BY 4.0<br> <strong>Contact:</strong> Masato Tanigawa <tanigawa@oita-u.ac.jp> (Department of Biophysics, Faculty of Medicine, Oita University)</p> <h3>Scope of this deposit</h3> <p>This Zenodo archive contains the <strong>code, input data, and analysis outputs (including supplementary tables)</strong> required to reproduce the numerical results of the manuscript. The manuscript text, cover letter, and typeset figures themselves are distributed through the journal (<em>PLOS Biology</em>) and are <strong>not</strong> included here to avoid duplication with the published record. Supplementary tables (Tables S1–S22) are included as tab-separated values and markdown so that reviewers can inspect and re-analyse them directly.</p> <h3>What is new in version 3 (2026-05-08, PLOS Biology submission)</h3> <p>Version 3 retains all version-2 content unchanged and adds five analytical extensions that strengthen the central protein-versus-regulation contrast:</p> <ol> <li><strong>DNA-repair gene set extension</strong> (<em>n</em> = 7 → 20 protein-level; <em>n</em> = 5 → 15 splice-site). Adds 13 <em>C. elegans</em>-anchored DNA-repair genes matched in size with the housekeeping (<em>n</em> = 20) and developmental (<em>n</em> = 20) categories. At matched native sample sizes, aging genes show ΔW = +0.434 versus DNA-repair ΔW = +0.091 — a five-fold contrast placing aging genes at the high-decoupling end of the category-level decoupling spectrum.</li> <li><strong>dN/dS extension</strong> (PAML yn00 on the 100-gene set). Aging <em>n</em> = 5 (Friedman <em>p</em> = 0.015, <em>W</em> = 0.444); DNA repair <em>n</em> = 10 (<em>p</em> = 6.8×10<sup>−4</sup>, <em>W</em> = 0.351); housekeeping <em>n</em> = 12 (<em>p</em> = 0.039, <em>W</em> = 0.273). dN/dS <em>W</em> lies between protein concordance and splice-site concordance, corroborating the protein-versus-regulation hierarchy at a third independent level.</li> <li><strong>TimeTree v5 sensitivity check</strong> (new Table S22). Pairwise Spearman ρ between clade-based and TimeTree v5 distance matrices: 0.825 (Nematoda), 0.846 (Insecta), 0.755 (Mammalia), all <em>p</em> < 10<sup>−39</sup>. Mantel-test results are essentially insensitive to the choice of distance convention.</li> <li><strong>Friedman robustness</strong> for the aging-subcategory investment profile (Table S20). Five independent checks confirm the baseline result (χ<sup>2</sup> = 9.80, <em>p</em> = 0.0074): bootstrap, leave-one-subcategory-out, leave-one-phylum-out, within-subcategory permutation, and trinucleotide-frequency control.</li> <li><strong>Quantitative reflections on minor methodological concerns</strong>. Splice-site composition: 96.11% canonical GT/AG, 3.89% non-canonical (dominated by GC-AG donors and scaffold-gap NN); 0.90% overall N content across 52,280 introns.</li> </ol> <p>The v3 manuscript is also reframed empirically (title and Abstract emphasise the <em>protein > gene-TPM > isoform > splice > intron</em> hierarchy as the central finding), with a Discussion paragraph reconciling the present cross-phylum decoupling result with the within-Mammalia per-event PSI–lifespan correlation of Jiang et al. (2025, <em>Nat. Commun.</em> 16:10317). These textual changes do not affect the data or the analytical pipeline.</p> <h3>What was already in versions 1 and 2</h3> <p><strong>Version 1</strong> (April 2026): cross-phylum splice-site analysis on 87 genes × 140 species (Nematoda + Insecta + Mammalia), Evo 2 7B embeddings, per-gene Mantel tests, category-level Kendall <em>W</em> comparison.</p> <p><strong>Version 2</strong> (May 2026): added (i) RNA-seq isoform quantification (Salmon, modENCODE + ENA + GTEx; three-indicator Kendall <em>W</em> hierarchy), (ii) species-level Bayesian network structure learning (pgmpy, 1,000-bootstrap edge stability), (iii) Phase B 5-phylum extension to Cnidaria + Echinodermata (1,659 orthologs, 12,322 splice sites, full Evo 2 embedding pipeline), and (iv) Phase C five-indicator hierarchy adding gene expression (TPM) and intron length.</p> <h3>Archive contents</h3> <p>Single top-level directory <code>zenodo_deposit_v3/</code> containing:</p> <ul> <li><code>README.md</code> — full documentation, changelog, and reproducibility instructions</li> <li><code>data/</code> — input data (gene categories, ortholog tables, splice-site sequences, Evo 2 embedding metadata)</li> <li><code>scripts/</code> — all analysis scripts (Python)</li> <li><code>results/</code> — per-analysis output directories (date-stamped)</li> <li><code>figures/</code> — supplementary tables (markdown and TSV)</li> </ul> <p>Archive size: 8.5 MB compressed (37.5 MB uncompressed), 198 files in 59 directories. See <code>README.md</code> for the complete file inventory and step-by-step reproduction instructions.</p> <h3>Verification</h3> <p>MD5: <code>5fb3bbcab749a664fdb1d77d174c6c95</code><br> SHA256: <code>9f829b3124606c3d7fde1807e88c7d19397148043dd20488cc002c8e412298f1</code></p>
