henhaobio
Sign in

bioRxivBioinformaticsPreprint

SenSASP: A Unified, Multi-Layer Database of Senescence and SASP Genes

Hao XuaniD, Yu HuangiD, Jiang BianiD

PDFDOICodeData
In-depth full-text digest available · Jump to digest ↓

Abstract

Research on cellular senescence and the senescence-associated secretory phenotype (SASP) draws on independently curated gene resources that differ in scope, identifiers, and update cycles, making cross-resource integration error-prone. We unified four widely used resources, CellAge, GenAge, the SenMayo signature, and the Reactome Cellular Senescence pathway, onto a single canonical identifier (the Ensembl gene ID) and enriched every gene with three annotation layers absent from all four inputs: cross-species conservation, tissue and cell-type expression, and high-confidence protein-protein interactions. Unification collapsed 1,460 summed source entries into 1,250 unique genes (210 redundant entries removed, 14.4%) while preserving full source provenance: 173 genes are corroborated by two or more resources and two (IL6, JUN) by all four. The three annotation layers reach 95.8%, 97.9%, and 93.0% of genes, with 89.4% annotated across all three. A 500-gene random sample of identifier mappings was validated against HGNC and Ensembl (98.0% exact match). The result, SenSASP, is a single, machine-readable, provenance-tracked database of harmonized identifiers and net-new functional context, illustrated here with a gene-prioritization score and a tissue-expression atlas. SenSASP is freely available at https://xuan13hao.github.io/sensasp/

In-depth summary

Claude Opus 5.5 · xhigh effort · 7 Nov 2026 · from the full text of version 2 (11 Sept 2026) · CC BY 4.0

Merging CellAge, GenAge, SenMayo and Reactome senescence genes onto Ensembl IDs yields 1,250 provenance-tracked genes, most also annotated with mouse/zebrafish orthologs, tissue expression and STRING interactions.

Four widely used gene resources on cellular senescence and the senescence-associated secretory phenotype (SASP) use different scopes, identifiers and update cadences, so combining them is manual and error-prone, and none reports conservation, expression or interaction context. The authors state that, to their knowledge, no prior resource joins these four on a shared identifier and links them to those three annotation layers.

Key results

  1. ShownMapping the four seed lists to Ensembl gene IDs and merging duplicates gave 1,250 unique genes, each with a record of its sources. Most genes come from only one resource.

    Inputs: 866 CellAge, 307 GenAge, 124 SenMayo and 163 Reactome entries (1,460 summed; 1,252 unique symbols). Merging removed 210 entries (14.4%). All 1,252 symbols mapped to Ensembl and 1,249 to UniProt; 5 outdated symbols were updated (e.g., ARNTL→BMAL1). 1,077 genes are single-source, 138 have two sources, 33 three, and 2 (IL6, JUN) all four. · Seed acquisition, Identifier harmonization, and…; Database summary statistics

  2. ShownEach of the three added annotation layers (conservation, expression, interactions) covers at least 93.0% of genes, and 89.4% of genes are annotated in all three. None of the four inputs provides these layers.

    Mouse ortholog 94.4%, zebrafish 83.6%, either 95.8%. GTEx v8 median expression across 54 tissues: 97.9%. Human Protein Atlas (HPA) tissue specificity: 97.9%. 93.0% of genes have at least one within-set STRING v12 interaction at combined score ≥700, giving 15,611 edges; TP53 is the top hub (degree 368). · Conservation, Expression, and Interaction layer; Database summary statistics

  3. ShownA manual check of identifier mappings against HGNC (HUGO Gene Nomenclature Committee) and Ensembl found few errors. Most mismatches were Ensembl IDs assigned to a duplicate copy of the gene on an alternate haplotype rather than on the primary assembly.

    Stratified sample of 500 of the 1,250 genes (every three- and four-source gene and every gene with a missing identifier, plus proportional draws of one- and two-source genes). HGNC, Ensembl, Entrez and UniProt fields matched exactly for 490/500 (98.0%; error 2.0%, 95% CI 1.1–3.6%). · Technical Validation

  4. SuggestedAn illustrative priority score combines source count, within-set STRING degree and ortholog identity. Its top 20 genes all have three or four sources, led by JUN, TP53 and IL6. The ranking barely changes after removing the overlap between source count and degree.

    Top-20 scores 100-142: JUN 142 (degree 155, 96.4% mouse identity), TP53 141 (degree 368), IL6 131. Source count and degree correlate (Spearman ρ = 0.33, Pearson r = 0.45); median degree is 12 in single-source vs 155.5 in four-source genes. Residualizing degree on source count left 19 of the top 20 unchanged. · Technical Validation; Example 1: SenSASP-guided multi-evidence gene…

  5. SuggestedClustering the GTEx expression of the top 15 priority genes across 15 ageing-relevant tissues gives groups the authors interpret as broadly expressed regulators, AP-1 genes, IL6, and cell-cycle-linked genes.

    Heatmap of log2(median TPM + 1), with rows and columns clustered. CTNNB1, STAT3 and RELA are broadly uniform. JUN and FOS peak in adipose and vascular tissue. IL6 is higher in whole blood than in solid tissues. ATM, CCNA2 and E2F1 are low overall and near-absent in brain and mature muscle. · Example 2: aging tissue expression landscape of…

Methods, data and reproducibility

Seed lists were downloaded from HAGR, MSigDB and the Reactome ContentService. Symbols were mapped with MyGene.info to HGNC, Ensembl, UniProt and Entrez IDs, then merged on the Ensembl gene ID with per-gene source records. Genes were annotated with Ensembl BioMart mouse/zebrafish orthologs, GTEx v8 and Human Protein Atlas expression, and STRING v12 interactions (score ≥700) restricted to the gene set.

  • CellAge senescence genes (Human Ageing Genomic Resources archive) · human (symbols mapped with human-restricted MyGene.info queries) · n = 866 genes
  • GenAge human ageing genes (Human Ageing Genomic Resources archive) · human · n = 307 genes
  • SenMayo senescence/SASP signature (SAUL_SEN_MAYO gene set, MSigDB) · human (symbols mapped with human-restricted MyGene.info queries) · n = 124 genes · M45803
  • Reactome Cellular Senescence pathway participants · human (R-HSA pathway; human-restricted mapping) · n = 163 genes · R-HSA-2559583
  • Mouse and zebrafish orthologs (type and percent identity) from Ensembl BioMart · Mus musculus, Danio rerio
  • Tissue expression: GTEx v8 median TPM plus Human Protein Atlas specificity, single-cell-type and subcellular location fields · human · n = 54 GTEx tissues
  • STRING v12 protein-protein interactions at combined score ≥700, restricted to edges within the gene set · human (Homo sapiens, taxon 9606) · n = 15,611 edges
  • SenSASP database output (JSON and CSV, one record per gene) with pipeline code · human · n = 1,250 genes · 10.5281/zenodo.21753677
Methods & tools
MyGene.info, Ensembl BioMart, GTEx v8 API, Human Protein Atlas API, STRING v12, Reactome ContentService, MSigDB, Human Ageing Genomic Resources (HAGR), HGNC, Ensembl
Reproducibility gaps
No software versions or dependency files are described. Seed-list release versions and download dates are not given. The pipeline queries live public APIs (MyGene.info, BioMart, GTEx, HPA, STRING), and releases are named only for GTEx v8 and STRING v12.
Limitations (6)
  • GenAge was kept whole instead of being cut to its senescence-specific evidence subset (about 25 genes), so users must filter it themselves using the per-gene source information. · stated by the authors
  • The priority-score weights are heuristic and not empirically calibrated, so the ranking should be read as hypothesis-generating rather than validated. · stated by the authors
  • Source count and within-set STRING degree are moderately correlated (multi-source genes tend to be better studied), so the score combines related rather than independent evidence. · stated by the authors
  • Non-coding entries without a clean protein mapping (e.g., TERC) have uneven conservation and interaction annotation. · stated by the authors
  • Ortholog, expression and interaction annotations depend on the assumptions and releases of Ensembl BioMart, GTEx v8, HPA and STRING v12. · stated by the authors
  • Technical validation covers identifier mapping only. The prioritization and tissue-expression examples are not tested against an independent set of known senescence drivers or any experimental data, and the tissue modules are read qualitatively from a clustered heatmap. · noted in this summary

Read this if you need one list of senescence and SASP genes with consistent identifiers, a record of which sources support each gene, and ortholog, tissue-expression and protein-interaction annotations for picking candidates.

Citations
0

Reproducibility ledger

Nobody has reported trying to reproduce this paper yet. Did you replicate an experiment or run the code? Your report saves the next lab a week.

Sign in to file a reproducibility report.

Related papers

Similarity Map
SelectedSame fieldCross-field
THISAvelar '19Saul '21Basisty '19Börner '26Tiwari '26Scanlan '22
66% semantic matchSystems Biology
Roberto A. Avelar, Javier Gómez Ortega, Robi Tăcutu+13 more·bioRxiv·Aug 23, 2019· 6 citations