Curation of the AMRFinderPlus databases: applications, functionality and impact

Abstract

Antimicrobial resistance (AMR) is a significant public health threat. Low-cost whole-genome sequencing, which is often used in surveillance programmes, provides an opportunity to assess AMR gene content in these genomes using in silico approaches. A variety of bioinformatic tools have been developed to identify these genomic elements. Most of those tools rely on reference databases of nucleotide or protein sequences and collections of models and rules for analysis. While the tools are critical for the identification of AMR genes, the databases themselves also provide significant utility for researchers, for applications ranging from sequence analysis to information about AMR phenotypes. Additionally, these databases can be evaluated by domain experts and others to ensure their accuracy. Here we describe how we curate the genes, point mutations and blast rules, and hidden Markov models used in NCBI’s AMRFinderPlus, along with the quality-control steps we take to ensure database quality. We also describe the web interfaces that display the full structure of the database and their newly developed cross-browser relationships. Then, using the Reference Gene Catalog as an example, we detail how the databases, rules and models are made publicly available, as well as how to access the software. In addition, as part of the Pathogen Detection system, we have analysed over 1 million publicly available genomes using AMRFinderPlus and its databases. We discuss how the computed analyses generated by those tools can be accessed through a web interface. Finally, we conclude with NCBI’s plans to make these databases accessible over the long-term.

Lelliottia amnigena recovered from the lung of a harbour porpoise, and comparative analyses with Lelliottia spp.

“…It is important to note that Bakta had annotated the bla LAQ gene on contig 81 of GFKo1’s genome (locus tag GFKo1_06635). Among its databases, Bakta uses the NCBI Antimicrobial Resistance Gene Finder (AMRFinderPlus) [37] to annotate AMR-associated genes in microbial genomes. In addition to a bla LAQ-1 -like gene, AMRFinderPlus predicted GFKo1 to encode vat (Vat family streptogramin A O -acetyltransferase; GFKo1_06890), catA (type A chloramphenicol O -acetyltransferase; GFKo1_12820) and oqxB (multidrug efflux RND transporter permease subunit OqxB; GFKo1_19950).…”

“…The genome sequence of GFKo1 was also searched for the allele of the chromosomal class C β-lactamase bla LAQ-1 (nucleotide accession MZ497396 [25]) using Geneious Prime v2023.0.1. Based on the result of the bla LAQ-1 search, AMRFinderPlus v3.11.4 (database version 2023-02-23.1) [37] and Bakta annotations were subsequently used for surveying AMR genes in genomes.…”

Molecular mechanisms of re-emerging chloramphenicol susceptibility in extended-spectrum beta-lactamase producing Enterobacterales

“…In our isolates, catB3Δ 443-633 is much more common than catB3 , and thus we investigated if this pseudogene is restricted to Malawi or common elsewhere. We queried “catB3” in NCBI‘s Microbial Browser for Identification of Genetic and Genomic Elements (Micro-BIGG-E) [35] and investigated 46,667 isolates (data download: 4/8/2023) with an annotated catB3 gene. Strikingly, 30,819 (66%) showed a coverage to the reference of 70%, which corresponds to the IS 26 truncated variant described here (Supplementary Fig.…”

Surveillance of carbapenem-resistant organisms using next-generation sequencing

“…Pathogen Detection provides AMR gene prediction for all submitted isolates in addition to SNP distances and phylogenetic trees for clustered isolates ( 22 , 24 ). An email notification alert was built to alert analysts when a submitted isolate is added into a SNP cluster on NCBI.…”