iPRESTO

iPRESTO detects and analyzes gene sub-clusters within biosynthetic gene clusters (BGCs) in microbial genomes to improve characterization of microbial specialized metabolism.

Key Features:

• Scalable sub-cluster detection: Performs unsupervised detection of sub-clusters within BGCs using topic modeling and statistical analysis of co-occurrence patterns among enzyme-coding protein families.
• Comprehensive dataset analysis: Has been applied at large scale to ~150,000 prokaryotic BGCs from antiSMASH-DB.
• Predictive and confirmatory performance: Predicts substructures for ~16% of antiSMASH-DB BGCs and confirms ~83% of experimentally characterized sub-clusters in MIBiG reference BGCs.
• Novel discovery and annotation: Detected sub-clusters have identified unannotated BGCs implicated in xenorhabdin and salbostatin biosynthesis and proposed a candidate BGC for akashin biosynthesis.
• Linking molecules to gene clusters: Links orphan metabolites to candidate gene clusters by correlating detected sub-clusters with MS/MS-derived Mass2Motifs, demonstrated across collections such as 145 actinobacteria.

Scientific Applications:

• Functional and structural annotation of BGCs: Predicts substructures and confirms known sub-clusters to support functional and structural annotation of microbial BGCs.
• Integration with metabolomics: Correlates BGC sub-clusters with MS/MS-derived Mass2Motifs to associate molecular substructures with biosynthetic loci.
• Natural product discovery and assignment: Supports discovery and characterization of novel bioactive compounds, exemplified by assignments for xenorhabdin, salbostatin, and a candidate for akashin.

Methodology:

Unsupervised topic modeling and statistical analysis of co-occurrence patterns among enzyme-coding protein families, correlation of detected sub-clusters with MS/MS-derived Mass2Motifs, benchmarking against MIBiG reference BGCs, and large-scale application to antiSMASH-DB (~150,000 BGCs).