AMR.watch logo

Overview

amr.watch is an interactive web application developed by the Centre for Genomic Pathogen Surveillance (CGPS) that enables monitoring of the WHO-defined priority bacterial pathogens and their associated AMR markers.

The application displays data inferred from high-quality genomes available in the public sequence archives, processed on an ongoing basis via our “Always-on” pipeline within Pathogen.watch.

amr.watch currently reports genome data for pathogens in the 2017 WHO priority pathogen list (WHO, 2017).

If you use amr.watch in a publication, please cite:

David S, Caballero JD, Couto N, Abudahab K, Fareed-Alikhan N, Yeats C, et al. amr.watch – monitoring antimicrobial resistance trends from global genomics data. bioRxiv. 2025 Apr 17;2025.04.17.649298

Data processing

Always-on pipeline

The amr.watch application uses genome assemblies generated and processed by our “Always-on” pipeline within Pathogen.watch (https://pathogen.watch). The pipeline retrieves all paired-end Illumina samples from the INSDC databases that have an available sampling date from 2010 onwards that is decodeable to at least the year, as well as a sampling location that is decodeable to at least the country level. Checks are performed to ensure the consistency and integrity of the data, including a requirement for two fastq files per sample and for sequencing runs to possess at least 20x coverage.

Sequence reads are assembled using a SPAdes workflow (https://gitlab.com/cgps/ghru/pipelines/assembly) and the quality of the resulting assemblies is assessed. Samples that do not meet our defined species-specific criteria (see publication (include hyperlink here)) are excluded. The Speciator tool within Pathogenwatch is used to verify the species of the assemblies, while SISTR (Yoshida et al. 2016) is additionally used to assign the serotype of the Salmonella enterica genomes. Genome assemblies with a species identification that is inconsistent with the recorded species in the ENA metadata are excluded. An exception here is made for E. coli and Shigella spp., for which the ENA metadata and Speciator assignments are permitted to be inconsistent, with the Speciator assignments used in subsequent processing.

Variant typing schemes

amr.watch displays the variants found among each pathogen using community-based schemes implemented in Pathogen.watch. For most of the species, these comprise multi-locus sequence typing (MLST) schemes from PubMLST (Jolley et al. 2018). For Acinetobacter baumannii, the “Pasteur” MLST scheme is used within amr.watch (rather than the “Oxford” scheme). Genomes are shown with a “Novel” sequence type (ST) within amr.watch either if the MLST profile is incomplete or if the profile has not been defined within the MLST database. In addition to MLST, we also use Global Pneumococcal Sequencing Cluster (GPSC) assignments for S. pneumoniae (Gladstone et al. 2019) and “clonal group” assignments from the LIN code nomenclature for K. pneumoniae (Hennart et al. 2022). For S. Typhi, we use the higher-resolution scheme, GenoTyphi (Dyson & Holt, 2021), rather than the MLST scheme.

Identification of mechanisms associated with AMR

Genes and mutations associated with particular antimicrobial classes are identified from the genome assemblies using AMRFinderPlus (database version 2021-12-21.1) (Feldgarden et al. 2021). A list of curated genes and mutations included for each pathogen and antimicrobial combination, obtained via a comprehensive literature review, is provided in the amr.watch publication (include hyperlink here). Complete matches to a gene are required for reporting of a mechanism within amr.watch. AMR mechanisms are identified for antimicrobial classes defined in the 2017 WHO priority pathogen list (WHO, 2017). We additionally report mechanisms for fluoroquinolone resistance in E. coli due its high global burden (Naghavi et al. 2024).

More details on the above processes can be found in the amr.watch publication (include hyperlink).

Using the application

Data limitations

As whole genome sequencing is increasingly being adopted into routine surveillance systems worldwide, shared public genomes offer a valuable resource for interrogating geotemporal trends around AMR. However, across the different pathogens, currently available public genomes lack broad geographic coverage and have largely been generated for specific research agendas. When using amr.watch, we therefore recommend that users remain vigilant to the data limitations and advise that additional investigations would be needed to confirm any identified trends observed from the data.

Navigating amr.watch

Genomes within amr.watch can be filtered on one or more criteria either using the top filter bar, the map or from within the right-hand panels. The number of genomes represented in amr.watch before and after the application of any filters is shown at the top of the page.

By default, the map panel colours countries by the number of genomes, after the application of filters (if any). Countries are coloured light grey if no genomes exist from the country in our present curated public collection of the relevant pathogen, or a darker grey if there are one or more genomes but which do not meet the filtering criteria. Users can filter the genomes by individual countries from the map. If one or more filters, other than the “Country” filter, are applied, the user can also opt to colour countries by the proportion of genomes that meet the filtering criteria.

The “Variants overview” panel shows the most frequent variants (up to twenty) among selected genomes, with the ability to toggle between different typing schemes if multiple schemes are available. Users can filter the genomes by individual variants from this panel, or from within the top panel if the variant of interest is not in the top twenty displayed.

The “Variant count over time” and the “Variant proportion over time” panels, which can be alternated, show the raw number of genomes and proportion of genomes belonging to all or selected variants per year.

The “AMR mechanism overview” panel shows the most frequent AMR mechanisms (up to twenty) for a particular antimicrobial class among selected genomes. Users can toggle between different antimicrobial classes if there are multiple classes. In these plots, genomes may be counted in more than one bar if they contain more than one AMR mechanism for the antimicrobial class. Users can filter the genomes by individual AMR mechanisms from this panel, or from within the top panel if the variant of interest is not in the top twenty displayed.

If there are no AMR mechanism filters applied, the “AMR mechanism count over time” and “AMR mechanism proportion over time” panels show the number/proportion of genomes with and without one or more mechanisms for the antimicrobial class that is selected in the above “AMR mechanism overview” panel. If one or more AMR mechanism filters are applied, the panels instead show the number/proportion of genomes with and without the selected mechanisms.

Saving/sharing a visualisation

Visualisations within amr.watch with a set of desired filters can be saved and/or shared onwards by clicking the “Share link” icon (top right) which saves a URL to your clipboard. A unique URL is generated for each visualisation that you would like to save/share.

Downloading data

Raw data can be downloaded by users in csv format using the “Download data” icon (top right). The downloaded data will represent all genomes shown in your current view and include a list of all genomes with ENA accession numbers, a link to their genome records, their variants and all AMR mechanisms found within the genomes for the WHO priority antimicrobial classes.

Funding

We are grateful for funding from the National Institute for Health Research (NIHR) and Bill & Melinda Gates Foundation.

Feedback

We would love to receive feedback on amr.watch and grateful to hear of any issues you are experiencing with the application. Please contact us at amrwatch@cgps.group

References

  1. Prioritization of pathogens to guide discovery, research and development of new antibiotics for drug-resistant bacterial infections, including tuberculosis [Internet]. [cited 2025 Mar 7]. Available from: https://www.who.int/publications/i/item/WHO-EMP-IAU-2017.12

  2. Yoshida CE, Kruczkiewicz P, Laing CR, Lingohr EJ, Gannon VPJ, Nash JHE, et al. The Salmonella In Silico Typing Resource (SISTR): An open web-accessible tool for rapidly typing and subtyping draft Salmonella genome assemblies. PLOS ONE. 2016 Jan 22;11(1):e0147101.

  3. Jolley KA, Bray JE, Maiden MCJ. Open-access bacterial population genomics: BIGSdb software, the PubMLST.org website and their applications. Wellcome Open Res. 2018;3:124.

  4. Gladstone RA, Lo SW, Lees JA, Croucher NJ, Tonder AJ van, Corander J, et al. International genomic definition of pneumococcal lineages, to contextualise disease, antibiotic resistance and vaccine impact. eBioMedicine. 2019 May 1;43:338–46.

  5. Hennart M, Guglielmini J, Bridel S, Maiden MCJ, Jolley KA, Criscuolo A, et al. A dual barcoding approach to bacterial strain nomenclature: Genomic taxonomy of Klebsiella pneumoniae strains. Mol Biol Evol. 2022 Jul 2;39(7):msac135.

  6. Dyson ZA, Holt KE. Five years of GenoTyphi: Updates to the global Salmonella Typhi Genotyping Framework. The Journal of Infectious Diseases. 2021 Dec 15;224(Supplement_7):S775–80.

  7. Feldgarden M, Brover V, Gonzalez-Escalona N, Frye JG, Haendiges J, Haft DH, et al. AMRFinderPlus and the reference gene catalog facilitate examination of the genomic links among antimicrobial resistance, stress response, and virulence. Sci Rep. 2021 Jun 16;11(1):12728.

  8. Naghavi M, Vollset SE, Ikuta KS, Swetschinski LR, Gray AP, Wool EE, et al. Global burden of bacterial antimicrobial resistance 1990–2021: a systematic analysis with forecasts to 2050. The Lancet. 2024 Sep 28;404(10459):1199–226.

AMR.watch is developed and maintained by the Centre for Genomic Pathogen Surveillance.