Amira: detection of single and multi-copy antimicrobial resistance genes directly from long reads using gene-space de Bruijn graphs

Abstract Accurate detection of antimicrobial resistance (AMR) genes is essential for AMR surveillance and genotypic prediction. Current approaches typically either depend on a correct genome assembly or use k-mers directly from the reads. However AMR genes are often multicopy and are enriched at repeat loci where genome assemblers tend to fail, and k-mer methods struggle to distinguish different copies. We present Amira, which detects AMR genes directly from long reads by building gene-space de Bruijn graphs, allowing precise association of reads with distinct AMR gene copies, enabling accurate assembly. Across simulations, curated samples, and public datasets, Amira improves copy-number recall and nucleotide accuracy compared with existing methods.