Use SEA-AD snRNA-seq (CELLxGENE Census slice, ~84 donors) to test whether MGnD microglia subtype abundance increases at earlier Braak stages than OPC/mature OL depletion. Approach: (1) Pull microglia and oligodendrocyte lineage cells from SEA-AD Census; (2) Leiden cluster at multiple resolutions post-Harmony integration; (3) Score subtypes against published marker sets (Olah, Sun, Mathys, Green et al. 2024); (4) Run scCODA/Dirichlet regression of subtype fractions on Braak stage, CERAD, Thal, age, sex, APOE; (5) Test temporal ordering with pseudotime or monotone regression of MG-DAM vs OPC fractions across Braak I–VI; (6) Cross-validate in ROSMAP Mathys 2023 cohort. Spatial validation: MERFISH co-localization of SPP1/GPNMB+ microglia voxels with MBP/PDGFRA+ oligo voxels and amyloid mask, by cortical layer. Confounders: donor age, PMI, sex, APOE4 dosage harmonized across cohorts.
Details
- disease
- alzheimer's disease
- target_ref
- UniProt:Q9NZC2 (TREM2), UniProt:P16234 (PDGFRA)
- primary_endpoint
- All three artifact IDs successfully read and at least one contains tabular donor-count data (not schema-only)
- identification_strategy
- observational
Raw fields (3)
- assay_spec
Read research_plan:18be54cc-ab1c-4f10-9e22-2b36354aa5ff and donor-harmonization datasets dataset:496e7ec9-9937-4852-a5ab-f02d5a392cf6 and dataset:2c9eb20c-5add-4ec5-a47d-13a3e513b454 via scidex.artifact.read; enumerate available columns (cohort, donor_id, Braak_stage, cell_type_label, microglia_subtype, OPC_subtype, n_donors_per_stratum, n_cells_per_stratum); apply >=10 donors per Braak-stratum gate; emit pass/blocker analysis_result artifact.; Enumeration table: cohort x Braak_stratum x cell_class (microglia_subtype, OPC) with n_donors and n_cells per cell; fields present/absent flagged explicitly — planned, not yet computed; {'kind': 'claim', 'text_template': 'Stratum X in cohort Y has only N donors (< 10); microglia-OPC coupling test is BLOCKED until stratum is augmented — OR — all strata meet >=10 donor threshold; proceed to coupling analysis'}; analysis_result artifact: 'AD cross-cohort feasibility gate v1' — pass/blocker decision with exact per-stratum donor counts and missing-field inventory; planned- hypothesis
In aged human cortex with advancing AD pathology (Braak III–VI), MGnD/DAM microglia expansion (marked by TREM2, SPP1, LPL, GPNMB) precedes and causally promotes oligodendrocyte loss (OPC depletion indexed by PDGFRA), mediated through myelin-phagocytic crosstalk and lipid-droplet accumulation in microglia. This ordering is detectable as a Braak-stage-dependent compositional shift in SEA-AD snRNA-seq, with MGnD fraction rising before mature OL/OPC fractions decline, and spatially co-localized with plaque burden in MERFISH cortical layers.
- kill_criteria
Abort or revise if: Artifacts 496e7ec9 and 2c9eb20c may contain only schema/metadata without actual donor-count rows — in that case the blocker is the artifact absence itself, not a threshold failure; Research plan 18be54cc may not specify stratum boundaries precisely; use Braak I-II / III-IV / V-VI as default bins if unspecified; Do not infer biology from field presence alone; donor-count sufficiency requires actual numeric values in the read result