BAGEL2

BAGEL2 identifies putative bacteriocin genes and annotates their genomic contexts in bacterial genomes, unfinished genome assemblies, and metagenomic datasets to enable discovery and characterization of antimicrobial peptides.

Key Features:

• Conserved domains and physical properties: Identifies potential bacteriocins using conserved sequence domains and characteristic physical properties of antimicrobial peptides.
• Genomic context analysis: Detects biosynthesis, transport, and immunity genes to evaluate the genomic neighborhoods of putative bacteriocin genes.
• High-throughput processing: Supports large-scale mining of complete and unfinished genomes as well as metagenomic samples.
• Parameter-free, class-specific mining: Performs class-specific bacteriocin mining without requiring user-defined parameters.
• Novel hidden Markov models (HMMs): Applies novel HMMs for prediction of bacteriocin sub-classes, interpreted using simple decision rules.
• Automated genetic context annotation: Annotates genetic contexts automatically using combinations of PFAM domains and databases of known context genes.
• Expert-validated database: Integrates an expert-validated database of bacteriocins for reference and comparison.
• Fine-tuned scoring system: Uses a scoring system refined with expert knowledge derived from screening all bacterial genomes available at NCBI.

Scientific Applications:

• Genome-scale bacteriocin discovery: Enables comprehensive mining of bacteriocins across diverse bacterial species and genome assemblies.
• Metagenomic discovery of novel antimicrobials: Facilitates identification of novel bacteriocins within metagenomic datasets.
• Functional and regulatory inference: Uses genomic context (biosynthesis, transport, immunity genes) to infer regulatory and functional aspects of bacteriocin production.
• Large-scale screening and curation: Supports large-scale surveys and curation workflows by combining predictive models with an expert-validated bacteriocin database and tuned scoring.

Methodology:

Identification based on conserved domains and physical properties; prediction of subclasses using novel HMMs interpreted by simple decision rules; automated annotation of genetic context using PFAM domain combinations and databases of known context genes; scoring refined by screening all bacterial genomes in NCBI.