PRISM 3

PRISM 3 predicts chemical structures of natural products from microbial genome sequences by detecting biosynthetic gene clusters and modeling candidate compounds as chemical graphs.

Key Features:

• Combinatorial structure prediction: Uses a combinatorial strategy to predict structures of nonribosomal peptides and polyketides and supports non-modular biosynthetic paradigms, representing natural product scaffolds as chemical graphs to enable prediction across diverse classes such as aminocoumarins, antimetabolites, bisindoles, and phosphonate natural products.
• Expanded cluster detection: Detects 22 distinct biosynthetic gene-cluster types, including 11 newly added cluster classes, broadening the range of detectable biosynthetic pathways.
• Enhanced sequence input and ORF detection: Improves handling of sequence input and open reading frame (ORF) detection to increase robustness and accuracy of downstream cluster identification and structure prediction.
• High-performance computing integration: Deployed on a 300-core server grid to increase throughput for analyses of large datasets.

Scientific Applications:

• Natural product discovery: Predicts chemical structures encoded by microbial genomes to support identification of novel natural products with pharmaceutical and industrial relevance.
• Biosynthetic linkage and prioritization: Links biosynthetic gene clusters to plausible compound scaffolds to prioritize candidate bioactive molecules and explore undercharacterized biosynthetic diversity.

Methodology:

Takes microbial genome sequences as input, detects open reading frames (ORFs) and identifies biosynthetic gene clusters, and applies a structure-prediction algorithm that models candidate natural products as chemical graphs and supports non-modular biosynthetic pathways.