Amira is an AMR gene detection tool designed to work directly from bacterial long read sequences. Amira makes it easy to reliably identify the AMR genes in a bacterial sample, reduces the time taken to get meaningful results and allows more accurate detection of AMR genes than assembly.
Amira leverages the full length of long read sequences to differentiate multi-copy genes by their local genomic context. This is done by first identifying the genes on each sequencing read and using the gene calls to construct a de Bruijn graph (DBG) in gene space. Following error correction, the reads containing different copies of multi-copy AMR genes can be clustered together based on their path in the graph, then assembled to obtain the nucleotide sequence.
Amira requires Python and three additional non-Python tools for optimal functionality:
- Python >=3.10,<3.13.
- Poetry to manage the Python dependencies.
- Pandora >=0.12.0 to identify the genes on each sequencing read.
- minimap2 for sequence alignment.
- samtools for processing alignments.
- racon for allele polishing.
- Jellyfish for cellular copy number estimation.
Follow these steps to install Amira and its dependencies.
Amira and its dependencies can be run through Singularity. A prebuilt singularity container can be found with each Release. Run this to build the container:
sudo singularity build amira.img Singularity.defYou can then run amira with the command below. NOTE: If you use the singularity container you do not need to specify the paths to any of the non-python dependencies. Amira will find them in the container automatically.
singularity run amira.img amira --helpAmira and all depdendencies, other than Pandora, can installed via conda by running the command below. NOTE: Amira requires Pandora v0.12.0 and this cannot be installed via conda at this time. You will need to build Pandora from source and specify the path to the binary with --pandora-path, or run Amira through a container.
conda install amira -c biocondaYou can then run amira with:
amira --helpAmira can be installed from PyPI. NOTE: You will need to install the non-Python dependencies separately if you opt for this method.
pip install amira-amrAmira can then be run with:
amira --helpOpen a terminal and run the following command to clone the repository and navigate into it:
git clone https://github.com/Danderson123/Amira && cd AmiraAmira’s dependencies are managed with Poetry. Install Poetry by running:
pip install poetryOnce Poetry is installed, use it to set up Amira’s python dependencies.
poetry installYou will need to install the non-Python dependencies separately if you opt for this method.
Amira requires Pandora, minimap2, racon and Jellyfish. Follow the links below for instructions on building binaries for each tool. The easiest way to run Pandora is through a binary that is precompiled for Linux, available from here.
- Pandora installation guide
- minimap2 installation guide
- samtools installation guide
- racon installation guide
- Jellyfish installation guide
After installation, make a note of the paths to these binaries as they will be required when running Amira.
Pandora uses species-specific reference pan-genomes (panRGs) to identify species-specific context genes on each sequencing read (see above for instructions to install Pandora). Amira uses these genes to differentiate the occurrences of multi-copy AMR genes in different genomic contexts. However, in theory, if you are only interested in detected AMR gene presence or absence, any reference panRG can be used as they all contain the same AMR genes.
Click the relevant link below to download a panRG to run Amira on your favorite bacterial species. If we do not currently support a species you are interested in then we are more than happy to build one, please let us know via a GitHub issue!
- Escherichia coli
- Klebsiella pneumoniae
- Enterococcus faecium
- Staphylococcus aureus
- Streptococcus pneumoniae
- ESKAPE pathogens + E. coli and Salmonella
Note: These panRGs can currently detect all acquired AMR genes in the NCBI Bacterial Antimicrobial Resistance Reference Gene Database as of 21st August 2024.
Once you have installed the Python dependencies and added Pandora, Racon, Minimap2 and Jellyfish to your $PATH, and downloaded the panRG for the bacterial species you interested in, Amira can be run with this command:
amira --reads <PATH TO READ FASTQ> --output <OUTPUT_DIR> --species <SPECIES> --panRG-path <PANRG PATH> --cores <CPUS>Amira can be run on assemblies using this command:
amira --assembly <PATH TO READ FASTA> --output <OUTPUT_DIR> --species <SPECIES> --panRG-path <PANRG PATH> --cores <CPUS>Default Amira assumes that you are running on single-isolate read sets of high depth, like you would use for whole-genome assembly, and AMR genes with a low relative-read depth are filtered from the output. Amira includes an experimental --meta mode that turns off all filtering of AMR genes. This can be useful if you are expecting AMR genes to occur on a very small number of reads. This can be run with:
amira --reads <PATH TO READ FASTQ> --output <OUTPUT_DIR> --species <SPECIES> --panRG-path <PANRG PATH> --cores <CPUS> --metaAmira can also be run on the output of Pandora directly, or from JSON files listing the genes and gene positions on each sequencing read.
After installing Pandora, you can call the genes on your sequencing reads using this command:
pandora map -t <THREADS> --min-gene-coverage-proportion 0.5 --max-covg 10000 -o pandora_map_output <PANRG PATH> <PATH TO READ FASTQ>Amira can then be run directly on the output of Pandora using this command:
amira --pandoraSam pandora_map_output/*.sam --pandoraConsensus pandora_map_output/pandora.consensus.fq.gz --panRG-path <PANRG PATH> --reads <PATH TO READ FASTQ> --output amira_output --species <SPECIES> --minimum-length-proportion 0.5 --maximum-length-proportion 1.5 --cores <CPUS> --racon-path <PATH TO RACON BINARY> --minimap2-path <PATH TO MINIMAP2 BINARY> --samtools-path <PATH TO SAMTOOLS BINARY>To run Amira from the JSON files, you can use this command:
amira --pandoraJSON <PATH TO GENE CALL JSON> --gene-positions <PATH TO GENE POSITION JSON> --pandoraConsensus <PATH TO PANDORA CONSENSUS FASTQ> --reads <PATH TO READ FASTQ> --output <OUTPUT DIRECTORY> --panRG-path <PANRG PATH> --species <SPECIES> --racon-path <PATH TO RACON BINARY> --minimap2-path <PATH TO MINIMAP2 BINARY> --cores <CPUS>
Some example JSON data can be downloaded from here. Amira can then be run using this command:
amira --pandoraJSON test_data/gene_calls_with_gene_filtering.json --gene-positions test_data/gene_positions_with_gene_filtering.json --pandoraConsensus test_data/pandora.consensus.fq.gz --reads test_data/SRR23044220_1.fastq.gz --output amira_output --species Escherichia_coli --panRG-path . --racon-path <PATH TO RACON BINARY> --minimap2-path <PATH TO MINIMAP2 BINARY> --samtools-path <PATH TO SAMTOOLS BINARY> --debug --cores <CPUS>
For additional options and configurations, run:
amira --help@article {Anderson2025.05.16.654303,
author = {Anderson, Daniel and Lima, Leandro and Le, Trieu and Judd, Louise and Wick, Ryan and Iqbal, Zamin},
title = {Amira: detection of AMR genes directly from long reads using gene-space de Bruijn graphs},
elocation-id = {2025.05.16.654303},
year = {2025},
doi = {10.1101/2025.05.16.654303},
publisher = {Cold Spring Harbor Laboratory},
abstract = {Accurate detection of antimicrobial resistance (AMR) genes is essential for the surveillance, epidemiology and genotypic prediction of AMR. This is typically done by generating an assembly from the sequencing reads of a bacterial isolate and running AMR gene detection tools on the assembly. However, despite advances in long-read sequencing that have greatly improved the quality and completeness of bacterial genome assemblies, assembly tools remain prone to large-scale errors caused by repeats in the genome, leading to inaccurate detection of AMR gene content and consequent impact on resistance prediction. In this work we present Amira, a tool to detect AMR genes directly from unassembled long-read sequencing data. Amira leverages the fact that multiple consecutive genes lie within a single read to construct gene-space de Bruijn graphs where the k-mer alphabet is the set of genes in the pan-genome of the species under study. Through this approach, the reads corresponding to different copies of AMR genes can be effectively separated based on the genomic context of the AMR genes, and used to infer the nucleotide sequence of each copy. Amira achieves significant improvements in genomic copy number recall and nucleotide accuracy, demonstrated through objective simulations and comparison with alternative read and assembly-based methods on samples with manually curated truth assemblies. Applied to a dataset of 32 Escherichia coli samples with diverse AMR gene content, Amira achieves a mean genomic-copy-number recall of 98.4\% with precision 97.9\% and nucleotide accuracy 99.9\%. Finally, we show that Amira consistently detects more true AMR genes across all E. coli, K. pneumoniae and E. faecium nanopore datasets from the ENA (n=8580, 2448 and 415 respectively) than an assembly-based approach.Competing Interest StatementThe authors have declared no competing interest.},
URL = {https://www.biorxiv.org/content/early/2025/06/06/2025.05.16.654303},
eprint = {https://www.biorxiv.org/content/early/2025/06/06/2025.05.16.654303.full.pdf},
journal = {bioRxiv}
}
If you’d like to contribute to Amira, please follow these steps:
- Fork the repository.
- Create a new branch for your feature or bugfix (git checkout -b feature-name).
- Commit your changes (git commit -m "Description of feature").
- Push to the branch (git push origin feature-name).
- Submit a pull request.
This project is licensed under the Apache License 2.0 License. See the LICENSE file for details.
For questions, feedback, or issues, please open an issue on GitHub or contact Daniel Anderson.