ayates/amr_portal

AMR Portal — Multi-Dataset Release (Phenotype + Genotype)

This repository contains multiple datasets from the EMBL-EBI AMR Portal, distributed in Apache Parquet format:

• phenotype.parquet – phenotypic antimicrobial susceptibility data
• genotype.parquet – AMR genes and mutations from in silico methods

All datasets are released under CC-BY-4.0.

Source documentation: https://www.ebi.ac.uk/amr/developers/

Dataset Summary

This dataset contains phenotypic antimicrobial resistance (AMR) data derived from antibiograms submitted to BioSamples from the CABBAGE data set. It provides curated experimental measurements such as MIC values, AMR gene annotation, and metadata describing geography, organism taxonomy, and antibiotics. The dataset is stored in Apache Parquet. This release corresponds to:

FTP location:
https://ftp.ebi.ac.uk/pub/databases/amr_portal/releases/2025-11/

Portal documentation:
https://www.ebi.ac.uk/amr/developers/

---

Dataset Contents

This dataset represents AMR phenotypic experimental outcomes collected from antibiograms.

Each row corresponds to an individual antibiotic test performed on an isolate.

Typical use cases include:

• AMR prediction
• Taxonomic/epidemiological AMR analysis
• Linking phenotypes to genomic or genotypic data

---

Data Schema

phenotype.parquet: AMR phenotypes

Field	Type	Nullable	Description
BioSample_ID	string	No	The unique identifier for the biological sample (e.g. SAMEA1028830)
SRA_accession	string	Yes	The SRA accession number
assembly_ID	string	Yes	The unique accession number of the genome assembly (e.g. GCA_001096525.1)
collection_year	int32	Yes	The year the sample was collected
ISO_country_code	string	Yes	The 3-letter ISO country code where the sample was collected (e.g. THA for Thailand)
host	string	Yes	The organism the sample was isolated from (e.g. Homo sapiens)
host_age	string	Yes	The age of the host (empty/NULL in the sample data, but should be a string to allow for various formats or NULLs)
host_sex	string	Yes	The sex of the host (empty/NULL in the sample data)
isolate	string	Yes	A unique identifier for the specific isolate (e.g. SMRU2695)
isolation_source	string	Yes	The specific anatomical source or environment the isolate came from (e.g. nasopharynx)
isolation_source_category	string	Yes	The general category of the isolation source (e.g. respiratory tract)
isolation_latitude	string	Yes	Geographic latitude for the sample
isolation_longitude	string	Yes	Geographic longitude for the sample
genus	string	No	The genus of the organism (e.g. Streptococcus)
organism	string	No	The full name of the organism (e.g. Streptococcus pneumoniae)
AMR_associated_publications	string	Yes	The PubMed ID of the publication associated with the data. Can be a set of values joined with a ;
Updated_phenotype_CLSI	string	Yes	The updated antimicrobial susceptibility testing (AST) phenotype based on CLSI standards (empty/NULL in the sample data)
Updated_phenotype_EUCAST	string	Yes	The updated AST phenotype based on EUCAST standards (empty/NULL in the sample data)
used_ECOFF	string	Yes	Indicates if the Epidemiological Cut-Off (ECOFF) was used (empty/NULL in the sample data)
database	string	Yes	Database of annotation
antibiotic_name	string	Yes	The name of the antibiotic tested (e.g. beta-lactams, trimethoprim-sulfamethoxazole)
ast_standard	string	Yes	The standard or guideline used for Antimicrobial Susceptibility Testing (e.g. CLSI, EUCAST)
laboratory_typing_method	string	Yes	The method used to test the antibiotic sensitivity (e.g. disk diffusion, E-test)
measurement	string	Yes	The raw measurement value, typically MIC or zone size (e.g. 2, 1, 0.5, 12/0.125).
measurement_sign	string	Yes	The sign indicating the nature of the measurement (e.g. '==' for exact value, or '>', '<')
measurement_units	string	Yes	The units for the measurement (e.g. mg/l)
platform	string	Yes	The platform used for analysis (empty/NULL in the sample data)
resistance_phenotype	string	Yes	The final result of the interpretation (e.g. susceptible, non-susceptible, resistant)
species	string	No	The species of the organism (e.g. Streptococcus pneumoniae)
antibiotic_ontology	string	Yes	An ontology ID for the antibiotic (e.g. ARO_3004024)
antibiotic_ontology_link	string	Yes	Link to the ontology resource for the ID
country	string	Yes	Full country name where the sample was collected from. Converted from ISO_country_code.
geographical_region	string	Yes	Geographical region as defined by UN M49. e.g Asia, Europe, Oceania, Africa or Americas.
geographical_subregion	string	Yes	Geographical subregion as defined by UN M49. e.g.Eastern Asia, Northern Europe.

genotype.parquet: AMR genotypes

Field	Type	Nullable	Description
BioSample_ID	string	No	The unique identifier for the biological sample (e.g., SAMEA1028830)
assembly_ID	string	No	The unique accession number of genome assembly (e.g., GCA_001096525.1)
genus	string	No	The genus of the organism (e.g., Streptococcus)
species	string	No	The species name of the organism
organism	string	No	The full name of the organism (e.g., Streptococcus pneumoniae)
isolate	string	Yes	Isolate information
taxon_id	int64	No	NCBI Taxonomy identifier of the organism
region	string	No	Name of a genomic region
region_start	int64	No	Start of the annotated gene
region_end	int64	No	End of the annotated gene
strand	string	No	Strand of the annotated gene. '+' indicates the forward strand, '-' indicates the reverse strand
_bin	int64	No	UCSC bin number for the genomic region. See UCSC's wiki for further details. Internal field
id	string	No	Identifier of the gene
gene_symbol	string	No	Symbol of the gene
amr_element_symbol	string	No	AMRFinderPlus assigned symbol for the AMR element
element_type	string	No	Broad type of AMR element. Normally set to AMR
element_subtype	string	No	Subtype of AMR element. Normally set to AMR
class	string	No	Overall class of AMR compound as given by AMRFinderPlus. Normally a broad representation of antibiotics
subclass	string	No	Subclass of AMR compound as given by AMRFinderPlus. Can also be set to the same as class
split_subclass	string	No	Subclass can represent multiple individual compounds separated by a '/'. This field contains the individual element of subclass.
antibiotic_name	string	Yes	Normalised name of the antibiotic tested (e.g., beta-lactams, trimethoprim-sulfamethoxazole)
antibiotic_ontology	string	Yes	Ontology ID for the antibiotic (e.g., ARO_3004024)
antibiotic_ontology_link	string	Yes	Link to ontology entry for the antibiotic
evidence_accession	string	Yes	Accession number for evidence supporting the predicted AMR resistance
evidence_type	string	Yes	Type of evidence supporting the predicted AMR resistance
evidence_link	string	Yes	Link to the evidence supporting the predicted AMR resistance
evidence_description	string	Yes	Evidence description supporting the predicted AMR resistance

---

How to Load the Dataset

Load phenotype data

from datasets import load_dataset

phenotype = load_dataset(
    "ayates/amr_portal",
    data_files="phenotype.parquet",
    split="train"
)

Load genotype data

genotype = load_dataset(
    "ayates/amr_portal",
    data_files="genotype.parquet",
    split="train"
)

Load everything at once

ds = load_dataset(
    "ayates/amr_portal",
    data_files={
        "phenotype": "phenotype.parquet",
        "genotype": "genotype.parquet"
    }
)

You can then access:

ds["phenotype"]
ds["genotype"]

License

Creative Commons Attribution 4.0 (CC-BY-4.0) https://creativecommons.org/licenses/by/4.0/

Citation

Please cite Dickens E et al. 2025 10.1101/2025.11.12.688105.

@article {Dickens2025.11.12.688105,
  author = {Dickens, Emily and Derelle, Romain and Beardmore, Robert and Suresh, Anita and Uplekar, Swapna and Yates, Andrew D and Keatley, Jon and Winterbottom, Andrea and Azov, Andrey G and El Houdaigui, Bilal and Ochkalova, Sofiia and Gurbich, Tatiana A and Shivalikanjli, Anu and Yordanova, Galabina and Lees, John A and Chindelevitch, Leonid},
  title = {A comprehensive AMR genotype-phenotype database (CABBAGE)},
  elocation-id = {2025.11.12.688105},
  year = {2025},
  doi = {10.1101/2025.11.12.688105},
  publisher = {Cold Spring Harbor Laboratory},
  URL = {https://www.biorxiv.org/content/early/2025/11/13/2025.11.12.688105},
  eprint = {https://www.biorxiv.org/content/early/2025/11/13/2025.11.12.688105.full.pdf},
  journal = {bioRxiv}
}