PRISM

PRISM predicts chemical structures of bacterial antibiotics from genome sequences to link genomic data to chemical structures and enable machine-learning prediction of biological activities.

Key Features:

• Genome-to-structure prediction (PRISM 4): Predicts chemical structures of bacterial antibiotics encoded in genomes.
• Coverage of antibiotic classes: Encompasses all classes of bacterial antibiotics currently utilized in clinical settings.
• Large-scale biosynthetic charting: Charts secondary metabolite biosynthesis across over 10,000 bacterial genomes sourced from cultured isolates and metagenomic datasets.
• Discovery output: Identified thousands of encoded antibiotics, expanding the repertoire of known antimicrobial agents.
• Supports machine learning: Produces predicted structures that enable development of machine-learning methods to forecast biological activities.

Scientific Applications:

• Linking sequence to structure: Connects microbial genome sequences to predicted secondary metabolite chemical structures.
• Antibiotic discovery from genomes: Enables identification of potential antibiotic compounds from cultured isolates and metagenomic datasets.
• Activity prediction: Provides input data for machine-learning models to forecast biological activities of predicted molecules.
• Comparative biosynthetic surveys: Facilitates large-scale surveys of secondary metabolite biosynthetic potential across bacterial genomes.

Methodology:

Predicts chemical structures of bacterial antibiotics from genome sequences; charts secondary metabolite biosynthesis across over 10,000 bacterial genomes from cultured isolates and metagenomic datasets; and produces predicted structures used to develop machine-learning methods to forecast biological activities.