Advancing viral genome classification: assesing the efficiency and accuracy of the alignment-free k-mer method in emerging pandemics

In January 2020, the World Health Organization officially recognized the novel coronavirus outbreak as a worldwide public health crisis, marking the onset of what would become the COVID-19 pandemic. Since then, extensive research efforts have been initiated to describe the virus, understand mutation patterns, transmission dynamics, and develop vaccines. Many of these studies require the classification of various virus strains, which is crucial for accurately characterizing the variants that emerged during the pandemic. However, classifying these strains requires methods for comparing genomic sequences, typically involving sequence alignment, a time-consuming process. In our study, we focused on assessing the accuracy and time efficiency of the k-mer method, which does not rely on sequence alignment but can enhance genomic comparisons. Using data from the National Center for Biotechnology Information SARS-CoV-2 website, we classified 17 complete genomes from different groups detected or emerging in Brazil, employing both alignment-based and k-mer approaches. An iterative prototype was developed in Shiny for classification analysis of SARS-CoV-2 virus sequences. Both methods yielded identical classifications, but the k-mer method outperformed significantly, being 97% faster. Therefore, we advocate for the use of the k-mer method in viral genome analysis, particularly during emerging pandemics. Its combination of speed and accuracy can greatly expedite responses to new viral threats.