Version history

{
  "disease": "immune aging / immunosenescence",
  "summary": "Cross-cohort characterization of age-associated CD8 T cell subsets in human blood, integrating Allen Immunology aging cohort, OneK1K, and Terekhova 2023 single-cell atlas data. Core reference confirmed: Terekhova et al., Immunity 2023 (DOI: 10.1016/j.immuni.2023.10.013; PMID: 37963457) — single-cell atlas of healthy human blood demonstrating age-related loss of NKG2C+GZMB−CD8+ memory T cells and accumulation of type 2 memory T cells across donors ages 20–89. Crossref validation complete (163 citations as of tick 81). Research plan targets: (1) harmonized taxonomy of CD8 exhaustion/memory subsets across cohorts, (2) pseudobulk DEG analysis with age + sex + CMV covariates, (3) compositional regression with donor random effects, (4) HALLMARK GSEA on age-associated DEGs, (5) sex-stratified and CMV-stratified trajectory analysis.",
  "assay_spec": "Tick 27 actions: (1) scidex.papers.fetch on Mogilenko 2021 DOI — planned output: n_donors, age_range, logFC, FDR for GZMK+CD8 TEM; (2) PubMed search for Terekhova 2023 — planned output: confirmed DOI + n_donors; (3) EuropePMC probe on OneK1K CMV metadata — planned output: CMV serostatus field presence/absence. Success criterion unchanged: ≥3 of 5 papers with all 6 gate-table fields non-null. Pseudobulk DE model spec held: limma-voom, covariates age_decade + sex + CMV_serostatus + batch, donor random effect. Kill criterion active: if OneK1K CMV serostatus confirmed absent, flag for cohort pivot.",
  "hypothesis": "Age-associated loss of NKG2C+GZMB−CD57+ CD8 memory T cells — documented in Terekhova 2023 — reflects a conserved blood immune-aging trajectory that (a) aligns with GZMK+ inflammaging signatures from Mogilenko 2021, (b) survives CMV-serostatus adjustment, and (c) diverges by sex across decade-binned age strata in cross-cohort analysis (Allen × OneK1K × Terekhova).",
  "target_ref": "gene:TOX, gene:PDCD1, gene:KLRG1, gene:B3GAT1, gene:CDKN1A",
  "kill_criteria": "NKG2C+CD57+ CD8 memory frequency effect size <0.2 Cohen's d across ≥2 independent cohorts after CMV adjustment; or cross-cohort label transfer F1 <0.6 for Terekhova subset taxonomy.",
  "timeline_weeks": 24,
  "primary_endpoint": "Age-decade effect size (Cohen's d) on NKG2C+CD57+GZMB− CD8 memory subset frequency per donor, adjusted for sex and CMV serostatus; FDR-corrected HALLMARK GSEA NES on pseudobulk age-DEGs.",
  "exclusion_criteria": "Donors with active infection, autoimmune diagnosis, immunosuppressive therapy, or HIV seropositivity. Datasets lacking CMV serostatus or sex metadata excluded from primary analysis; included as sensitivity cohort.",
  "inclusion_criteria": "Human donors with PBMC single-cell RNA-seq and available age, sex, and CMV serostatus metadata; age range 20–89; datasets with ≥20 donors and ≥500 CD8 T cells per donor.",
  "statistical_methods": "Pseudobulk aggregation per donor per CD8 subset; limma-voom with age (continuous and decade-binned) + sex + CMV covariates; compositional regression (scCODA / miloR-style) with donor random effect; HALLMARK GSEA (fgsea) on age-ranked DEGs; cross-cohort reference mapping via Symphony or scArches before pooled analysis.",
  "identification_strategy": "observational"
}