Characterize cross-cohort microglia subtype taxonomy reconciliation (SEA-AD vs ROSMAP vs Mathys 2023) and assess microglia–oligodendrocyte coupling along AD Braak trajectory. Primary open question: do SPP1/GPNMB/LPL-high MGnD-like states expand before or after oligodendrocyte loss, and does this ordering differ by APOE genotype or sex? Integration strategy: Harmony on donor-level covariates, Leiden clustering at resolutions 0.3–1.2, marker crosswalk against Olah/Sun/Mathys taxonomies, scCODA compositional regression on Braak I-VI. Kill criterion: subtype ARI < 0.4 against reference taxonomy after batch correction.
Details
- disease
- alzheimers
- target_ref
- dataset:SEA-AD:snRNAseq:84donors
- kill_criteria
- ARI < 0.4 between SEA-AD MGnD cluster and Mathys 2023 equivalent after Harmony correction; or n < 50 donors with both Braak stage and APOE genotype annotated across cohorts
- timeline_weeks
- 24
- exclusion_criteria
- Donors lacking Braak stage annotation; libraries with doublet rate > 10%; donors with concurrent non-AD neurodegenerative diagnoses.
- inclusion_criteria
- Donors with both snRNA-seq and neuropathology staging (Braak I–VI, CERAD, Thal); microglia nuclei count ≥ 50 per donor after QC; age ≥ 60.
- identification_strategy
- observational
Raw fields (4)
- assay_spec
Tick 46 evidence sweep — four calls: (1) EuropePMC: SPP1/GPNMB microglia + OPC/oligodendrocyte loss in human AD cortex snRNA-seq, reframed from myelin degeneration to OPC precursor loss; (2) PubMed: TREM2 microglia + lipid homeostasis + myelin phagocytosis + oligodendrocyte in human AD brain 2022-2025, capturing lipid-stress mechanistic framing; (3) EuropePMC: resilience + preserved cognition + homeostatic microglia (P2RY12/TMEM119) in human AD snRNA-seq cortex; (4) CrossRef DOI verify: 10.1038/s41593-024-01774-5 (candidate DOI for Gabitto/Travaglini SEA-AD Nature Neuroscience 2024 flagship). Planned output: if mg-oligo coupling papers returned, add as evidence_link to hypothesis cc20ae25. If SEA-AD flagship DOI confirms, add citation provenance to research_plan.
- hypothesis
MGnD-like microglia (SPP1+, GPNMB+, TREM2-low) expand in Braak III-IV cortex coincident with or preceding oligodendrocyte lineage loss, and this coupling is stronger in APOE4 carriers and female donors, implicating a lipid-stress–driven neuroimmune cascade as an early driver of cortical demyelination in AD.
- primary_endpoint
scCODA posterior probability that MGnD-like microglia fraction increases monotonically across Braak I-VI in SEA-AD snRNA-seq; secondary: Pearson correlation between oligo lineage fraction and MGnD fraction per donor stratified by APOE genotype and sex
- statistical_methods
Dirichlet-multinomial regression (scCODA) for subtype composition vs Braak/CERAD/Thal; Pearson and Spearman correlation for oligo vs microglia fraction per donor; permutation test for APOE4 × sex interaction on MGnD fraction; silhouette score and ARI-to-reference for cluster stability