# C1 reproduction benchmark check v2 — 2026-10-30
## Conclusion: PASS (39/39)

## Background
- The v1 anchor (AJOG 2024 positive marker genes, direction assumed to be up) failed 1/9 — root cause:
  the AJOG abstract does not state gene directions; "expected=up" was an unverified analyst assumption,
  i.e. an anchor-definition error, not a pipeline error (PDCA Act: redefine the anchor).
- v2 anchor: the mitonuclear preprint (Research Square, rs-10060346) independently confirmed with DESeq2
  on the **same dataset** that 13 mtDNA-encoded respiratory-chain genes are significantly down-regulated
  in GDM placenta (log2FC ≈ -1.4, adj.P<1e-9).

## Result of this pipeline (GSE249311, DESeq2, ~subtype, |LFC|>=0.58 & padj<0.05)
- A1   vs Control: mt 13/13 down, LFC -1.41 to -1.66
- A2   vs Control: mt 13/13 down, LFC -1.23 to -1.41
- T2DM vs Control: mt 13/13 down, LFC -0.90 to -1.10
- Directional concordance 39/39; LFC magnitudes match the preprint → metadata mapping and pipeline
  correctness independently validated.

## Notes
1. The AJOG anchor genes (CSH1/PER1/FOXO1/EGFR/SOD3/DOCK5/SOGA1) are mostly down-regulated in this
   pipeline's GDMA1 contrast — to be re-evaluated after checking the qPCR directions in the AJOG
   original article; the SOD3 down-regulation is consistent with the GDM oxidative-stress literature,
   suggesting the original direction may indeed be down or subtype-specific.
2. Low-library sample NCS61 (Control, 1.6M reads): retention to be decided by sensitivity analysis.
3. DEG summary (|LFC|>=0.58, padj<0.05): A1 3950 (up 1368/down 2582), A2 1306 (666/640),
   T2DM 3717 (1586/2131).
