GRAViTy-V2: a grounded viral taxonomy application

Taxonomic classification of viruses is essential for understanding their evolution and therefore their distribution, host interactions and pathogenic mechanisms. Classification methodologies usually rely on comparison of aligned sequence motifs in conserved genes, by genome organisation and gene complements, and at lower taxonomic ranks such as genus and species, through genome sequence identities. Building on our previous classification framework based on a novel whole-genome analysis method, we here describe Genome Relationships Applied to Viral Taxonomy Version 2 (GRAViTy-V2), which encompasses a greatly expanded range of features and numerous optimisations, packaged as an application that may be used as an alignment-free general-purpose virus classification tool. Using 28 datasets derived from the International Society on Taxonomy of Viruses 2022 taxonomy proposals, GRAViTy-V2 output was compared against human expert-curated classifications used for assignments in the 2023 round of ICTV taxonomy changes. GRAViTy-V2 produced taxonomies equivalent to manually-curated versions down to the family level and in almost all cases, to genus and species levels. However, discrepancies with our results primarily arose through various human and automated sequence annotation errors and erroneous annotations of coding sequences used in their original classification. Analysis times ranged from 1–506 min (median 3.59) on datasets with 17–1004 genomes and mean genome length of 3,000–1,000,000 bases, on a standard consumer-grade laptop. We discuss how the output from GRAViTY-V2 outputs allows for a full analysis of why taxonomic classifications were proposed, the value of the program for quality control of genetic comparisons, and how to optimise the speed of classification through proper use of GRAViTy-V2’s workflow management system.