SIMMER

SIMMER predicts microbial species and enzymatic activities in the human gut microbiome that catalyze chemical transformations of drugs, foods, and toxins.

Key Features:

• Chemical similarity algorithms: Uses advanced chemical similarity algorithms to model and compare substrates and reactions.
• Protein similarity algorithms: Employs protein similarity and sequence similarity searches to map enzymes to predicted chemical transformations.
• Joint species–enzyme prediction: Predicts both the microbial species and their corresponding enzymatic activities responsible for specific gut chemical reactions.
• Benchmarking: Recapitulated 88 known drug transformations and validated predictions against external datasets.
• Experimental support: Predictions were supported by in vitro validation using methotrexate.
• Accuracy improvement: Demonstrates improved accuracy relative to previous models that used simplistic chemical representations and sequence search schemes.

Scientific Applications:

• Drug metabolism mapping: Identifies enzymes and species responsible for known and putative drug transformations in the human gut, exemplified by methotrexate.
• Metabolism of ingested compounds: Infers microbial enzymatic capabilities for the metabolism of foods and environmental toxins.
• Candidate prioritization: Prioritizes candidate enzymes and species for targeted experimental validation and discovery of novel bacterial enzymes affecting ingested compounds.

Methodology:

In silico predictions combine advanced chemical similarity algorithms with protein similarity and sequence similarity searches, and predictions are benchmarked against external datasets.