Query AMR data
Latest data release yyyy-mm
How to use the portal About the portal

About the portal

The AMR (antimicrobial resistance) Portal is a collaboration between EMBL-EBI research and service teams alongside Imperial College London to deliver a new way to access and explore AMR phenotype and genotype annotations through a user-friendly interface. Data within the portal is taken from the MRC project CABBAGE (Comprehensive Assessment of Bacterial-Based AMR prediction from Genotypes) and represents the largest public AMR dataset in a reconciled, uniform format. AMR represents a major growing concern to public health predicted to cause over 2M deaths per year by 2050, with the Global South facing a disproportionate burden of AMR. It increases the risk of untreatable bacterial infections whilst increasing the chance of complications from essential medical procedures from childbirth to transplants.

CABBAGE

The Comprehensive Assessment of Bacterial-Based Antimicrobial resistance prediction from GEnotypes (CABBAGE) project aims to provide a unified AMR phenotype–genotype database for the WHO Bacterial Priority Pathogens (currently following the 2017 WHO list). The database integrates genotype–phenotype information extracted from public resources (PATRIC, NCBI antibiograms, NARMS, NDARO, CDC, COMPARE, PathogenWatch, Microreact, and PubMLST) as well as large datasets obtained from the literature. Isolate metadata such as collection year and country were also retrieved from the BioSample database when missing from the original sources.

Each entry in the database is defined by an isolate BioSample ID and an antibiotic name, and includes either a measurement value (MIC or disk diffusion zone diameter), an SIR phenotype, or both. Additional SIR phenotype inferences were performed using the latest CLSI and EUCAST breakpoint values through the AMR R package (v3.0.0).

To ensure comparability across sources, several data standardisation steps were applied. For example, antibiotic abbreviations were converted to full names using a single naming scheme. Further curation steps improved data quality, such as removing erroneous measurements and correcting AST method labels. A de-duplication process was finally implemented to merge duplicate entries and remove unresolved phenotype conflicts originating from different sources.

Please note that not all entries have been manually curated, only those that were clearly inconsistent or conflicted. CABBAGE will continue to be refined as new data are added. The current version of the database contains 1.7M entries, representing 171k isolates.

More information on CABBAGE can be found in the preprint Dickens E et al. 2025.

AllTheBacteria

Assemblies used for genotyping were generated by the AllTheBacteria collaboration. The assembly pipeline, uses Shovill (v1.1.0), and are processed with a simple pipeline to download reads, run Shovill, and remove contigs matching the human genome (using MUMmer v4.0.0rc1). Full methods are in the preprint Hunt M et al. 2025.

AMR Phenotypes

CABBAGE has catalogued an immense collection of antibiograms in a standardised format which has been processed by the EMBL-EBI teams to a single consistent large data format. This includes linkage to the Antibiotic Resistance Ontology (ARO) ontology or ChEBI (Chemical Entities of Biological Interest) resource. As part of this work we are brokering these normalised antibiograms back to BioSamples.

AMR Genotypes

AMR Genotypes is a resource recording the detection and annotation of resistance-associated genes and mutations. All annotated genomes have been processed using the mettannotator workflow from EMBL-EBI's MGnify team and generates an exhaustive annotation of prokaryotic genomes using existing tools. We have currently run AMRFinderPlus from NCBI against these genomes. UniProt's UniFIRE functional annotation system will provide additional annotations. Results from these tools are normalised according to the same rules as our phenotype data.

Portal contributors

The AMR portal is a collaboration between the following Imperial College London and EMBL-EBI teams and groups:

  • Chindelevitch Team (IC)
  • European Nucleotide Archive (EMBL-EBI)
  • Genomics Technology Infrastructure (EMBL-EBI)
  • Lees Group (EMBL-EBI)
  • Microbiome Informatics (EMBL-EBI)
  • Protein Function Development (EMBL-EBI)
  • Protein Sequence Resources (EMBL-EBI)
  • Samples, Phenotypes, and Ontologies (EMBL-EBI)

Future goals

This project represents phase one of our efforts. Future developments may include

  • Advanced querying including search by IDs
  • Expanding AMR profiles to those from isolate genomes in culture collections
  • Expanding AMR profiles from text mining the literature
  • Submission flows allowing third parties to submit antibiograms and for these data to flow into this portal
  • Further standardisation of the antibiogram formats
  • Display of isolate genomes in the Ensembl resource including integration of AMR functional annotation