Review Papers Claim by Claim

Key strengths (what this paper contributes)

• Large, focused dataset: 4,713 complete ESKAPE genomes screened and 1,790 AmpC proteins catalogued — useful baseline for comparative genomics and surveillance (Pandey et al. 2025 dataset & methods).
• Replication of known biology: ADC chromosomal ubiquity in A. baumannii and PDC prevalence in P. aeruginosa are consistent with prior literature (strengthens internal validity of their pipeline) (PDC/AmpC in P. aeruginosa (prior evidence)).

Major limitations & potential biases

• Sampling bias: analysis restricted to 'complete genomes' from NCBI as of Sept 2023 — publicly available assemblies overrepresent clinical/epidemic lineages and certain geographies; genomes submitted to NCBI are not a random sample of global diversity (paper acknowledges NCBI download but does not fully quantify sampling biases).
• 100% identity BLAST filter: Authors kept only sequences at 100% identity to BLDB entries. That strict threshold reduces false positives but may exclude legitimate natural variants (novel alleles), underestimating diversity and novel plasmid-mediated AmpC emergence.
• Phenotype missing: No linkage shown between genomic AmpC variants and antimicrobial susceptibility phenotypes (MICs) or expression/regulatory mutations (e.g., ampR/ampD). This leaves the functional impact of many identified substitutions unresolved.
• Representatives selection for BEAST: choice of 'top MSA sequence' as representative per enzyme group (first MSA row) can bias tree timing and branch-length inferences — representativeness not demonstrated quantitatively.
• Plasmid localization calls: assigning plasmid vs chromosomal location from GenBank annotations can be confounded by misassembled replicons or integrated plasmid sequences; short-read assemblies and misannotations can mislead localization statistics.

Where the evidence is strongest

1. Presence/absence signal: the pipeline and pHMM + validation with BLDB appear to robustly detect canonical AmpC sequences in the sampled assemblies (authors claim 100% specificity during GenBank annotation cross-checks) (Pandey et al. 2025 pHMM specificity claim).
2. Structural comparisons: using high-identity PDB representatives for eight groups gives credible RMSD comparisons and supports sequence-based clustering (table of PDB IDs and RMSDs provided in the manuscript).

Where confidence is lower

• Evolutionary timing: time‑stamped BEAST on one representative per group (selected as first MSA sequence) provides a rough chronology but is underpowered for reliable divergence dating across highly recombining, horizontally-mobile genes.
• Functional inference: motif analysis (SXXK, YXN, KTG) is useful, but absence/alteration in PIB (noted) does not by itself quantify catalytic changes — biochemical assays or MIC associations are needed to support phenotype claims.

Conclusions & practical implications

Pandey et al. 2025 produce a high-value catalog and cross‑species map of AmpC diversity and replicon-association that is useful as a reference for surveillance, annotation, and primer/assay design; however, the work should be interpreted as a genomic/structural atlas, not as proof of clinical resistance phenotypes for each novel sequence. Key operational takeaways: (1) surveillance should track ADC in A. baumannii and plasmid-borne CMY/DHA/ACT in Enterobacterales; (2) strict identity filtering is conservative but will miss novel/worrying variants — future updates should relax thresholds and incorporate functional assays and expression/regulatory context.

All core claims above are grounded in the preprint dataset and figures (Pandey et al. 2025 (preprint)).