Augmenting Bacterial Similarity Measures Using a Graph-Based Genome Representation

Abstract

Relationships between bacterial taxa are traditionally defined using 16S rRNA nucleotide similarity or average nucleotide identity. Improvements in sequencing technology provides additional pairwise information on genome sequences, which may provide valuable information on genomic relationships. Mapping orthologous gene locations between genome pairs, known as synteny, is typically implemented in the discovery of new species and has not been systematically applied to bacterial genomes. Using a dataset of 378 bacterial genomes, we developed and tested a new measure of synteny similarity between a pair of genomes, which was scaled onto 16S rRNA distance using covariance matrices. Based on the input gene functions used (i.e., core, antibiotic resistance, and virulence), we observed varying topological arrangements of bacterial relationship networks by applying (1) complete linkage hierarchical clustering and (2) KNN graph structures to syntenic-scaled 16S data. Our metric improved clustering quality comparatively to state-of-the-art ANI metrics while preserving clustering assignments for the highest similarity relationships. Our findings indicate that syntenic relationships provide more granular and interpretable relationships for within-genera taxa compared to pairwise similarity measures, particularly in functional contexts. O_FIG O_LINKSMALLFIG WIDTH=200 HEIGHT=82 SRC="FIGDIR/small/588602v1_ufig1.gif" ALT="Figure 1"> View larger version (25K): org.highwire.dtl.DTLVardef@1bf81edorg.highwire.dtl.DTLVardef@1eaf18dorg.highwire.dtl.DTLVardef@198912dorg.highwire.dtl.DTLVardef@1633ee5_HPS_FORMAT_FIGEXP M_FIG GRAPHICAL ABSTRACT.This method uses a novel measure based on shared synteny between bacterial genomes to scale 16S rRNA distance values. Created with BioRender.com. C_FIG