COMPARISON · ANTIMICROBIAL RESISTANCE BIOINFORMATICS

Detecting antibiotic resistance genes: which tool should you use?

AMRFinderPlus, ResFinder, PointFinder, CARD/RGI, ABRicate and KMA are the most widely used tools for finding antibiotic resistance genes (ARGs) and resistance mutations in sequenced bacterial genomes. They do not take the same input data, do not rely on the same reference databases, and not all of them detect mutations. This comparison summarises their differences, based on their publications and official documentation.

By Rachid Merah, veterinarian by training, author of MEGAM ARG Detection · Published 2 November 2026. MEGAM uses several of these tools; the information below comes from their official sources, cited at the end of the page.

Comparison table

Features change between versions: check each tool’s documentation before an analysis. Last checked: 2 November 2026.

Choosing according to your situation

You have an assembled genome

AMRFinderPlus, RGI and ABRicate work on assembled sequences. AMRFinderPlus relies on the NCBI reference database and, if you specify the species, also searches for known point mutations. RGI uses CARD’s detection models, including variant models. ABRicate is fast and lets you compare several databases (ResFinder, CARD, NCBI…), but it only looks for acquired genes.

You only have sequencing reads

KMA aligns raw reads directly against redundant databases such as ResFinder’s; ResFinder uses it to process reads. RGI offers the bwt mode, which aligns FASTQ reads against CARD reference sequences. AMRFinderPlus and ABRicate require an assembly first.

You are looking for chromosomal mutations

Resistance to quinolones, colistin or rifampicin often arises from mutations in chromosomal genes rather than from the acquisition of a gene. PointFinder was designed for this; AMRFinderPlus searches for them in supported species when the species is specified; RGI covers them with its variant models. ABRicate does not detect them.

You want a predicted antibiogram

ResFinder 4.0 links each determinant to a phenotype and provides an in silico antibiogram for some species. Its validation covered Escherichia coli, Salmonella, Campylobacter jejuni, Enterococcus faecium, Enterococcus faecalis and Staphylococcus aureus.

You are analysing a metagenome

This comparison is mainly about bacterial isolates. For metagenomes, RGI in bwt mode and dedicated tools such as DeepARG, a deep learning approach designed for metagenomic data, are better suited.

You do not want to use the command line

The ResFinder and CARD web services run in a browser but require uploading your sequences to their servers. Otherwise, most of these tools are installed on Linux or macOS and run from the command line. A local desktop application avoids both.

Why tools give different results

Reference databases differ in content and naming; identity and coverage thresholds differ; an assembly and a read-based analysis do not always see the same genes; and sequencing quality matters too. In an inter-laboratory study, nine teams received the same sequencing data from clinical isolates: they found different numbers of resistance genes and different gene variants. The authors note that, had these results been used to guide treatment, a different antibiotic would have been recommended for each isolate by at least one team (Doyle et al., 2020).

In practice:

• always report the tool, its version and the database version;
• check the identity and coverage of important detections;
• compare several tools or databases when needed;
• remember that a detected gene is not enough to conclude clinical resistance: only an antimicrobial susceptibility test measures susceptibility.

References

1. Feldgarden M et al. AMRFinderPlus and the Reference Gene Catalog facilitate examination of the genomic links among antimicrobial resistance, stress response, and virulence. Sci Rep. 2021;11:12728. doi:10.1038/s41598-021-91456-0
2. Bortolaia V et al. ResFinder 4.0 for predictions of phenotypes from genotypes. J Antimicrob Chemother. 2020;75(12):3491-3500. doi:10.1093/jac/dkaa345
3. Zankari E et al. PointFinder: a novel web tool for WGS-based detection of antimicrobial resistance associated with chromosomal point mutations in bacterial pathogens. J Antimicrob Chemother. 2017;72(10):2764-2768. doi:10.1093/jac/dkx217
4. Alcock BP et al. CARD 2023: expanded curation, support for machine learning, and resistome prediction at the Comprehensive Antibiotic Resistance Database. Nucleic Acids Res. 2023;51(D1):D690-D699. doi:10.1093/nar/gkac920
5. Clausen PTLC, Aarestrup FM, Lund O. Rapid and precise alignment of raw reads against redundant databases with KMA. BMC Bioinformatics. 2018;19:307. doi:10.1186/s12859-018-2336-6
6. Seemann T. ABRicate: mass screening of contigs for antimicrobial and virulence genes. github.com/tseemann/abricate
7. Arango-Argoty G et al. DeepARG: a deep learning approach for predicting antibiotic resistance genes from metagenomic data. Microbiome. 2018;6:23. doi:10.1186/s40168-018-0401-z
8. Doyle RM et al. Discordant bioinformatic predictions of antimicrobial resistance from whole-genome sequencing data of bacterial isolates: an inter-laboratory study. Microb Genom. 2020;6(2). doi:10.1099/mgen.0.000335

Official documentation: AMRFinderPlus · ResFinder · RGI · CARD · KMA