Augmenting bacterial similarity measures using a graph-based genome representation.

Area of Science:

• Genomics
• Bioinformatics
• Microbial Ecology

Background:

• Bacterial relationships are traditionally assessed using 16S rRNA nucleotide similarity or average nucleotide identity (ANI).
• Advancements in sequencing technology enable the use of genome-wide data, such as synteny, for bacterial classification.
• Synteny, the mapping of orthologous gene locations, has not been systematically applied to bacterial genome relationship analysis.

Purpose of the Study:

• To develop and test a novel synteny similarity measure for bacterial genomes.
• To integrate synteny information with 16S rRNA data for enhanced bacterial relationship analysis.
• To explore graph-based modeling of bacterial genomes for new analytical approaches.

Main Methods:

• A dataset of 378 bacterial genomes was analyzed.
• A new synteny similarity metric was developed and scaled onto 16S rRNA distance using covariance matrices.
• Complete linkage hierarchical clustering and K-nearest neighbor graph structures were applied to synteny-scaled data, considering core, antibiotic resistance, and virulence genes.

Main Results:

• The synteny similarity metric improved clustering quality compared to state-of-the-art ANI metrics.
• The approach preserved clustering assignments for highly similar relationships.
• Varying topological arrangements of bacterial relationship networks were observed based on gene function inputs.

Conclusions:

• Syntenic relationships offer more granular and interpretable insights into within-genera bacterial taxa compared to pairwise similarity measures.
• This functional and synteny-based layer enhances bacterial identification and genome clustering.
• Graph structure modeling of bacterial genomes opens new avenues for genomic analysis of bacteria and their close relatives.