Naïve Bayes Classifiers and accompanying dataset for Pseudomonas syringae isolate characterization

Area of Science:

• Plant pathology
• Microbial genomics
• Bioinformatics

Background:

• The Pseudomonas syringae species complex (PSSC) comprises diverse plant pathogens threatening global food security.
• Current phylogenetic identification methods are hindered by complex and evolving taxonomy, complicating routine diagnostics.
• Existing amplicon sequencing methods lack efficient and accurate classification tools for PSSC.

Purpose of the Study:

• To develop a suite of Naïve Bayes classifiers for accurate PSSC identification using PCR primer data.
• To establish a method for translating classifier outputs into traditional taxonomic ranks (species, phylogroup, pathovar).
• To enable prediction of virulence factor repertoires for PSSC isolates.

Main Methods:

• Generated in-silico amplicon data from 2,161 published PSSC genomes.
• Employed the life identification number (LIN) hierarchical clustering algorithm for taxonomic analysis.
• Developed five Naïve Bayes classifiers tailored to widely used PCR primer sets for PSSC.

Main Results:

• The Naïve Bayes classifiers demonstrate efficient and accurate classification of PSSC isolates based on amplicon data.
• A comprehensive dataset was created for mapping classifier results to established taxonomic nomenclature.
• The study provides a framework for predicting the virulence factor repertoires of PSSC.

Conclusions:

• The developed Naïve Bayes classifiers offer an efficient solution for accurate PSSC identification, overcoming taxonomic challenges.
• This approach enhances the practical application of genomic data for routine diagnostics and epidemiological surveillance.
• The ability to predict virulence factors aids in assessing the potential threat of emerging PSSC lineages to agriculture.