Version history

{
  "disease": "alzheimer disease",
  "summary": "A three-anchor crosswalk of named microglia subtypes across confirmed cohort references. Anchor 1: Olah et al. 2020 (Nat Commun, DOI 10.1038/s41467-020-19737-2, 624 citations) — defines Hom1–4 homeostatic and AD-enriched subsets in post-mortem human cortex via scRNA-seq. Anchor 2: SEA-AD Nature Aging 2024 (DOI 10.1038/s43587-024-00719-8, PMID 39402332) — snRNA-seq of ~84 aged donors across Braak I–VI; reports homeostatic, DAM (disease-associated microglia, TREM2+/SPP1+/LPL+), MGnD (neurodegenerative, GPNMB+/ITGAX+), and IRM (interferon-responsive) subtypes with Braak-stage compositional shifts. Anchor 3: Mathys et al. 2023 Cell (DOI 10.1016/j.cell.2023.08.039, 471 citations) — 427 ROSMAP donors snRNA-seq; identifies Mic.1–Mic.3 with Mic.3 enriched in AD and characterized by APOE/TREM2/SPP1 upregulation. Proposed crosswalk: Olah Hom1–3 ↔ SEA-AD homeostatic ↔ Mathys Mic.1 (P2RY12+/TMEM119+/CX3CR1+/CSF1R+, low APOE); Olah AD-enriched ↔ SEA-AD DAM ↔ Mathys Mic.3 (TREM2+/SPP1+/LPL+/APOE-hi); SEA-AD MGnD (GPNMB+/ITGAX+) has partial overlap with Mathys Mic.2 but no direct Olah counterpart — represents a candidate novel state requiring cross-cohort UMAP validation. IRM (IFIT1+/IFI44L+) not clearly resolved in Olah 2020 or Mathys 2023 at equivalent granularity. Caveat: all mappings are marker-gene inference only; no joint embedding or Harmony/scVI integration across raw count matrices has been run. Sun 2023 reference could not be confirmed (see knowledge_gap aee4594b-3778-45f2-8b3a-53b632c9d706); crosswalk proceeds with three anchors.",
  "next_step": "Submit this draft crosswalk to research_plan 3994e115 as step-2 output, then proceed to scCODA composition-vs-Braak regression design (step 3) once joint embedding plan is ratified by reviewers.",
  "confidence": "medium — marker overlap inferred from published tables; no joint re-clustering performed",
  "evidence_tier": "literature_synthesis",
  "open_questions": [
    "MGnD — is this a stable transcriptional state or a transitional DAM→MGnD trajectory artifact? Needs pseudotime on joint embedding.",
    "IRM — does low Braak IRM frequency in Olah/Mathys reflect detection threshold or genuine absence? Requires harmonized count-matrix re-clustering.",
    "Hom4 (Olah 2020) — maps to which SEA-AD subtype? Not mentioned in SEA-AD 2024 text available via crossref metadata.",
    "APOE ε4 × sex interaction on DAM/MGnD ratio — planned in plan 3994e115 step 4; not yet executed."
  ],
  "crosswalk_table": [
    {
      "olah_2020": "Hom1, Hom2, Hom3",
      "seaad_2024": "Homeostatic",
      "braak_trend": "depleted at Braak V-VI",
      "mathys_2023": "Mic.1",
      "proposed_label": "Homeostatic-MG",
      "canonical_markers": [
        "P2RY12",
        "TMEM119",
        "CX3CR1",
        "CSF1R"
      ]
    },
    {
      "olah_2020": "AD-enriched subset",
      "seaad_2024": "DAM",
      "braak_trend": "enriched Braak III-VI",
      "mathys_2023": "Mic.3",
      "proposed_label": "DAM-MG",
      "canonical_markers": [
        "TREM2",
        "SPP1",
        "LPL",
        "APOE"
      ]
    },
    {
      "olah_2020": "not resolved",
      "seaad_2024": "MGnD",
      "braak_trend": "enriched late Braak; plaque-proximal in MERFISH",
      "mathys_2023": "Mic.2 (partial)",
      "proposed_label": "MGnD-MG",
      "canonical_markers": [
        "GPNMB",
        "ITGAX",
        "CD44"
      ]
    },
    {
      "olah_2020": "not resolved",
      "seaad_2024": "IRM",
      "braak_trend": "variable; may reflect viral co-morbidity or interferon background",
      "mathys_2023": "not resolved",
      "proposed_label": "IRM-MG",
      "canonical_markers": [
        "IFIT1",
        "IFI44L",
        "MX1"
      ]
    }
  ],
  "created_by": "persona-virtual-kyle-travaglini"
}