SanntiS

SanntiS identifies biosynthetic gene clusters (BGCs) using machine learning trained on InterPro signatures to detect a broad spectrum of BGC classes in genomic and metagenomic datasets.

Key Features:

• Machine learning-based BGC detection: SanntiS employs neural networks trained on InterPro signatures for SMBGC annotation.
• Genomic and metagenomic compatibility: SanntiS achieves high precision and recall in both genomic and metagenomic datasets, including assemblies with incomplete BGC genes.
• Broad BGC class coverage: SanntiS effectively captures a wide range of BGCs beyond those overrepresented in existing databases.
• Contextual data integration: BGC predictions include associated contextual data from diverse biomes.

Scientific Applications:

• Metagenomic BGC mining: Application to MGnify metagenomic assemblies generated 1.9 million BGC predictions from diverse environments.
• Novel natural product discovery: SanntiS detected a novel antimicrobial peptide that was subsequently experimentally validated.
• Comparative novelty assessment: BGC predictions from metagenomic datasets demonstrate significant novelty compared to equivalent isolate genome datasets.

Methodology:

SanntiS uses neural networks trained on InterPro signatures for SMBGC annotation.