Review Papers Claim by Claim

Visual highlights (evidence-backed)

• Scope & scale: FARME aggregates 11,057 DNA sequences (≈19.9 Mb) and 26,253 predicted proteins from 20 functional metagenomics projects — all drawn from environmental (uncultured) sources and annotated with BLAST and HMMs FARME DB core numbers
• Annotation strategy: sequences were BLASTed against GenBank and annotated with Pfam & Resfams HMMs via HMMER; HMMs were used with "trusted cutoffs" and overlapping models resolved by bitscore FARME methods (BLAST + HMMER)HMMER3 methods
• Unique contribution: FARME focuses exclusively on functionally validated ARG elements from environmental metagenomics (including flanking regulatory/mobile genetic elements), complementing clinically-derived AR databases that under-represent environmental functional hits FARME unique focus

Critical appraisal (strengths, limitations, blind spots)

• Strength — empirical grounding: FARME contains sequences that were functionally selected (clones conferring resistance in surrogate hosts), so entries represent experimentally validated phenotypes rather than purely in silico predictions; authors emphasize functional metagenomics as the data source Functional selection evidence
• Strength — genomic neighborhood: FARME deliberately stores flanking predicted proteins and HMM-annotated regulator/MGE features, enabling investigation of mobilization potential and regulatory contexts that are absent from many clinical AR databases Genomic neighborhood data
• Limitation — coverage & project selection bias: FARME aggregates sequences from 20 published functional metagenomics projects available in GenBank (selection depends on what teams published and deposited); thus FARME's sample of environmental niches and geographic coverage is constrained by existing published studies (authors note soil and gut dominate) Project selection & biome skew
• Limitation — functional context dependence: Functional selection in a surrogate host (commonly E. coli) may not recapitulate activity in native hosts — a clone that confers resistance in E. coli may use host-specific expression, folding, or interactions; conversely, some genuine environmental ARGs may be missed if they fail to express or function in the surrogate system (known limitation of functional metagenomics) Functional metagenomics caveat
• Limitation — annotation dependence: Many FARME proteins align to GenBank entries (authors report ~69% >80% identity), but functional annotation still relies on HMMs and BLAST; where homology is low (authors report ~1,334 proteins <80% similar to both GenBank and ARDB) these may be novel but require biochemical/structural follow-up to prove mechanism Sequence similarity statistics
• Reproducibility & transparency: Methods (GenBank retrieval, MetaGeneMark predictions, BLAST thresholds, HMM databases/versions, and website schema) are described — the authors provide the FARME website for access (link supplied) and list versions (Pfam v29, Resfams v1.1, HMMER 3.1b2), aiding reproducibility Reproducibility statements

What would refute or alter FARME's central claims?

1. Systematic failure of function: demonstrating that a substantial fraction of FARME sequences fail to confer resistance when re-tested in multiple, physiologically-relevant hosts would weaken the assertion that entries represent broadly functional ARGs Functional re-test importance
2. Annotation mismatch: if HMM predictions and BLAST-based annotations systematically mis-assign AR activity (false positives), then downstream interpretations (e.g. AR HMM counts) would change; orthogonal biochemical/structural validation would overturn certain classifications.

Author reviews (quick links)

Selected citations used in this review

• FARME DB paper
• HMMER3 methods
• BLAST foundational methods