Answers are grounded in this paper and cite their sources.
Genomic studies of antimicrobial resistance (AMR) rely on several independently curated reference databases, which describe the same genes under different names, in different structures and with different classifications. This makes results obtained with different databases difficult to compare. ARGeement maps the entries of AMRFinderPlus, the Comprehensive Antibiotic Resistance Database (CARD) and ResFinder/DisinFinder to each other by protein sequence, independently of their names. It characterises every link by its relation type, difference count and coverage, tracks the database versions, and produces a documented, immutable mapping for every release. The 17,920 entries of the release described here encode 17,959 protein sequences, corresponding to 9,337 unique proteins. For 64.4% of the cross-database entry pairs that encode the same protein, the names differ even when letter case is ignored, and 447 pairs show a genuine naming conflict. Conversely, 2.0% of the pairs with the same allele name do not encode the same protein. For synthetic variants created with known modifications, the original protein was among the best hits in 99.6% of the cases. In 100 complete RefSeq genomes annotated with AMRFinderPlus, the Resistance Gene Identifier (RGI) and ResFinder, the gene names agreed at 55.3%-75.6% of the loci called by both tools of a pair, whereas after mapping the two reference proteins were identical, near-identical or length variants at 89.6%-94.2%. These results show that name-based comparison is unreliable even for identical sequences, and that a sequence-based, versioned mapping is needed.
No takes yet. Share an insight, caveat, or question.
Vujić et al. (2026) studied this question.
Synapse has enriched 8 closely related papers on similar clinical questions. Consider them for comparative context: