Abstract
Accurate taxonomic assignment of bacterial genomes is essential for identifying novel taxa and for stable classification to enable robust comparison between studies. Bacterial genus delineation relies on multiple lines of evidence, including phylogenetic trees and metrics like the percentage of conserved proteins (POCP). POCP is widely used, but requires benchmarking in terms of both, computation and accuracy. We used 2,358,466 pairwise comparisons of proteomes derived from 4,767 genomes across 35 families to systematically assess POCP calculation and percentage of conserved proteins with unique matches (POCPu) which considers unique matches only. Both methods are 20x faster than the reference BLASTP when using the very-sensitive setting of DIAMOND. However, POCPu differentiates better within-genus from between-genera values, which improves bacterial genus assignment. This work facilitates comparative analysis of an increasingly larger number of genomes, providing a reliable metric to support genus delineation. The findings suggest that specific POCPu thresholds deviating from the reference 50% value are needed for certain families.