A novel core genome clustering method proposed for population structure analysis and genotyping of bacterial genomes

Core genome multilocus sequence typing (cgMLST) has emerged as an effective method for population structure analysis and genotyping. However, it exhibits limited discriminative power for bacterial genomes that are highly recombinogenic or display low genetic diversity. To address this limitation, we propose cgKNV, a novel core genome clustering method based on the k-mer natural vector, which overcomes the resolution constraint and maintains computational efficiency. We evaluate the utility of cgKNV using three datasets with different genetic diversities: 346 highly recombinogenic Enterococcus faecium for population structure analysis, 583 genetically monomorphic Bacillus anthracis for genotyping and linkage inference, and a comprehensive collection of 1786 Mycobacterium abscessus for dominant circulating clone identification. Our results demonstrate that cgKNV provides more precise descriptions of population structure and higher discriminatory powers in genotyping compared to cgMLST. As a threshold-free clustering method, cgKNV classified isolates within the established population structure and effectively grouped bacterial genomes across varying levels of genome diversity. Furthermore, cgKNV is highly scalable that supports easy integration of new isolates, making it as an efficient tool to perform large-scale bacterial genome studies.