#methodological rigourCentral analyses use non-independent observations
Several imaging tests treat hundreds of cells as inferential observations, while the post-mortem χ² analysis treats 763 cells from 20 cases as independent. These analyses do not account for clustering within wells, experiments, flies, or human cases.
↳ Figs. 1b,e,h, 4b–f, and 8
#reportingReported statistics conflict with interpretations
Fig. 6c contains an incompatible F/p pair, while the direct rescue comparisons in Figs. 4f and 6k are non-significant despite stronger narrative claims. These discrepancies create a material Internal Coherence concern.
↳ Figs. 4f and 6c,k legends; corresponding Results sections
#contributionHuman disease burden is over-attributed
Cross-sectional morphology and k-means clusters identify cells with compatible features but cannot show that karyoptosis caused their degeneration or quantify a causal 18–20% share. The case-level sampling structure is also absent from the inferential model.
↳ Results, Fig. 8 and final paragraph before Discussion