Paper Review — Verify Claims Fast

1) What the paper did (facts from the text)

• Collected snRNA-seq and snATAC-seq data across 17 species (239,512 transcriptomic nuclei; 201,147 ATAC nuclei) and performed Visium spatial transcriptomics (regular and HD) in key species to map liver cell types and zonation Data collection and integration summary
• Inferred that hepatocyte zonation (portal-central metabolic division) is present across mammals and absent in sampled non-mammalian vertebrates (bird and bichir) and links this to WNT/RSPO central-VE expression and a conserved TCF7L2-centered eGRN Zonation origin and WNT RSPO TCF7L2 claim
• Reported rapid evolutionary turnover of zonated genes (many gains/losses across branches) yet conservation of core metabolic zonation functions (amino acid catabolism portal; ammonia detox central) Zonation turnover and conserved pathways
• Highlighted human lineage shifts in cell type specificity (example CD36 moving to endothelial expression), with supporting snATAC chromatin-accessibility changes and identification of human-specific CREs enriched for lipid pathways Human specific CD36 and CRE changes

2) Strengths

• Unprecedented taxonomic breadth and multiomic depth (17 species, transcriptome + chromatin + spatial), raw sample counts and accession numbers provided for reproducibility Data availability and accessions
• Appropriate integration of modalities and careful QC: doublet detection, UMI/intron fraction filtering for nuclei, LSI + Harmony for ATAC, Seurat CCA for RNA integration — methods are standard and state-of-the-art Processing and QC pipeline
• Cross-validation between snRNA, snATAC, and Visium HD spatial data gives orthogonal evidence for zonation and human-specific shifts (good multi-modality corroboration) Cross-modality validation

3) Key weaknesses, blindspots and places where claims exceed evidence

1. Sampling heterogeneity and cell recovery bias: central endothelial cells were underrepresented in several species and the authors explicitly merged low-count central endothelial cells with central LSECs for ligand analysis; this weakens claims about conserved ligand expression in species with low cell recovery Central endothelial sampling limitation
2. Temporal and circadian confounders: authors acknowledge rhythmic expression and included both sexes, but timing of sampling across species may bias zonation calls (27/127 human-mouse species-specific genes are rhythmic in mouse), and limited circadian control reduces confidence that all interspecies differences are spatial rather than temporal Temporal rhythm caveat
3. Ancestral state reconstruction fragility: the fuzzy c-means + phylogenetic averaging approach is reasonable, but ancestral inference depends on exclusion thresholds (correlation >0.4, SD <1.5). Rapid turnover claims require sensitivity tests to these thresholds; the paper provides criteria but limited sensitivity analyses are shown in main text Ancestral inference method and thresholds
4. Functional causality not proven for WNT/RSPO->TCF7L2 across non-model species: eGRN and motif/ChIP comparisons implicate TCF7L2, and mouse ChIP support is used, but direct functional perturbation (eg endothelial WNT deletion or TCF7L2 loss) across multiple species is absent; evolutionary causal claim is plausible but not yet functionally demonstrated across mammals TCF7L2 evidence and functional gap
5. CRE orthology annotations and LiftOver limitations: mapping human CREs across great apes relies on HAL liftover HALPER and LiftOver conversions between assemblies (PanTro6 to PanTro5 etc), which can introduce false negatives for CRE presence in non-human apes; authors restricted to robust peaks but residual assembly/alignability biases may inflate 'human-specific' CRE calls CRE orthology and liftOver caveat

4) Reproducibility and data access

The authors provide raw accessions for snRNAseq snATACseq Visium data (E-MTAB series listed) plus code repository and a web app; this greatly improves reproducibility. Key dependencies and parameters are specified (eg ArchR addGroupCoverages MACS2 settings, ArchR addPeak2GeneLinks corCutOff 0.45), facilitating replication Data access and pipeline parameters

5) How strong are the central claims and what would falsify them?

• Claim: Zonation is a mammalian innovation driven by WNT/RSPO from central vein endothelial cells and TCF7L2 — Evidence: multi-species snRNA/snATAC + spatial co-localization + eGRN nominate TCF7L2; Strength: moderate-to-strong but not causal. Falsification: demonstrating identical, spatially-patterned WNT/RSPO signaling and hepatocyte zonation in non-mammalian vertebrates (eg chicken, bichir) using controlled spatial assays, or knocking-out WNT/RSPO or TCF7L2 in mammals and showing zonation persists would disprove causality Primary causal claim and falsification

6) Practical next steps and experiments to strengthen conclusions

1. Targeted functional perturbations: Conditional endothelial deletion of Rspo/Wnt ligands or hepatocyte TCF7L2 loss in a non-rodent mammal (eg sheep) combined with Visium HD spatial readout to test causal role in zonation.
2. Controlled sampling for circadian state: sample matched circadian timepoints across species to exclude rhythmicity artifacts for species-specific zonation genes.
3. Orthology benchmarking: test HALPER/Liftover-based CRE absence calls by resequencing targeted orthologous regions (ATAC-PCR) in great apes to reduce assembly/alignability false negatives.

7) Quick reproducibility checklist for a lab re-running main analyses

• Download E-MTAB accessions listed and verify sample meta matches paper (species, replicate, sex, tissue region).
• Re-run nucleus filtering thresholds (UMI, mitochondrial, intronic fraction) and doublet removal using scrublet; compare cell-type proportions.
• Run ArchR pseudobulks and MACS2 peak calling with parameters as reported and reproduce CRE sets; compute PhastCons/GERP constraints as described.
• Recompute fuzzy c-means clustering and ancestral inference using paper thresholds then test sensitivity to correlation cutoff and SD cutoff.

8) Direct citations to the source evidence (verbatim extracts used above)

• Paper main abstract statement
• Data sizes and accessions

9) Final balanced judgment

The manuscript is a major data resource and a convincing descriptive atlas showing that mammalian hepatocyte zonation is widespread and that WNT/RSPO plus TCF7L2 are strong candidate mediators. The strongest contributions are (1) taxonomic breadth and matched chromatin+transcriptome+spatial evidence and (2) the identification of lineage-specific regulatory changes (eg CD36). The main limitations are sampling heterogeneity across species, incomplete functional perturbation evidence for cross-species causality, and potential assembly/alignability biases in CRE orthology calling. Overall confidence in the descriptive claims is high; causal evolutionary mechanism claims are plausible but need targeted functional validation.