Structure of the Enterobacter pan-genome is revealed using machine learning

Area of Science:

• Microbiology
• Genomics
• Bioinformatics

Background:

• Enterobacter species are significant nosocomial pathogens known for antimicrobial resistance.
• Understanding the genomic diversity of Enterobacter is crucial for studying its evolution and resistance mechanisms.
• Publicly available Enterobacter genomes provide an opportunity to analyze its pangenome structure.

Purpose of the Study:

• To reveal the structure of the Enterobacter pangenome by analyzing gene distribution across species and subspecies.
• To identify major modes of gene inheritance, both lineage-associated and horizontal, within the Enterobacter genus.
• To enable robust phylogenetic and functional classification of Enterobacter genomes.

Main Methods:

• Analysis of 777 high-quality complete Enterobacter genomes using a pangenome matrix.
• Decomposition of the accessory genome using non-negative matrix factorization (NMF) to identify gene groups (Phylons).
• Classification of an additional 2,291 fragmented genome sequences based on the defined pangenome structure.

Main Results:

• Defined 31 Phylons, comprising 21 lineage-associated gene sets and 10 Phylons associated with mobile genetic elements (plasmids and chromosomal DNA).
• Demonstrated that the identified pangenome structure is consistent with the classification of fragmented genome sequences.
• Enabled pangenome-wide mapping of genetic traits, including motility, biosynthesis, antimicrobial resistance, and virulence factors.

Conclusions:

• The 31 identified Phylons define the Enterobacter pangenome structure, revealing dominant modes of inheritance.
• This NMF-driven classification enhances the understanding of Enterobacter evolution and resistance potential.
• The study provides a robust classification of Enterobacter isolates, clarifying taxonomic ambiguities and identifying differential genetic traits.