RODEO

RODEO identifies biosynthetic gene clusters (BGCs) and predicts ribosomally synthesized and post-translationally modified peptide (RiPP) precursor peptides, with emphasis on lasso peptide discovery, using hidden Markov model analysis, heuristic scoring, and machine learning.

Key Features:

• Biosynthetic gene cluster identification: Detects and evaluates BGCs associated with RiPP production.
• RiPP precursor peptide prediction: Predicts RiPP precursor peptides, including hypervariable lasso peptide precursors.
• Hidden Markov model analysis: Uses HMM-based methods to identify sequence motifs and domain architectures relevant to RiPP biosynthesis.
• Heuristic scoring: Applies heuristic scoring to rank candidate ORFs and biosynthetic elements.
• Machine learning techniques: Incorporates machine learning to improve prediction accuracy for precursor peptides and cluster annotation.
• Lasso peptide space mapping: Maps lasso peptide sequence space and identifies over 1,300 distinct compounds.
• Candidate prioritization: Prioritizes candidate peptides based on predicted structural novelty, enabling selection for further study.
• Characterized lasso peptides: Resulted in characterization of six unique lasso peptides, including one with a handcuff-like topology and one bearing a rare citrulline modification in bacteria.

Scientific Applications:

• RiPP discovery: Enables discovery and annotation of novel RiPP natural products, particularly lasso peptides.
• Genome mining: Facilitates large-scale genome mining for BGCs encoding RiPPs across genomic datasets.
• Prioritization for experimental validation: Ranks candidates for downstream biochemical characterization based on predicted structural novelty.
• Dataset expansion for natural product research: Expands searchable collections of lasso peptides and supports broader BGC-driven investigations and re-engineering efforts.

Methodology:

RODEO applies hidden Markov model-based analysis, heuristic scoring, and machine learning to map lasso peptide sequence space, identify over 1,300 distinct compounds, and prioritize candidate peptides by predicted structural novelty.