GutBugDB: A Web Resource to Predict the Human Gut Microbiome-Mediated Biotransformation of Biotic and Xenobiotic Molecules

In recent years, the significant role of the human gut microbiota (HGM) in the metabolism of drugs and nutrients has gradually been recognized. The gut microbiota not only affects the bioavailability of orally administered drugs but also participates in the biotransformation of drugs and bioactive molecules through its metabolic enzymes, thereby influencing the pharmacokinetics and pharmacodynamics of drugs. However, due to the complexity of the gut microbiota and the variability among individuals, determining the specific contributions of particular microbes to the metabolism of drugs and nutrients remains a significant challenge. To address this issue, researchers have developed GutBugDB, an open-access digital repository aimed at predicting the human gut microbiome-mediated biotransformation of biotic and xenobiotic molecules.

Source of the Paper

The study was conducted by Usha Longwani, Ashok K. Sharma, Aditya S. Malwe, Shubham K. Jaiswal, and Vineet K. Sharma from the Metabiosys Lab at the Indian Institute of Science Education and Research (IISER) Bhopal. The paper was published in 2025 in the journal Gut Microbiome under the title GutBugDB: A Web Resource to Predict the Human Gut Microbiome-Mediated Biotransformation of Biotic and Xenobiotic Molecules.

Research Process

1. Data Collection and Classification

The study began by collecting 1,439 FDA-approved drugs and nutrients from the U.S. Food and Drug Administration (FDA) database, supplemented by a literature review. These drugs were classified into 14 pharmacological categories, such as autonomic nervous system drugs, respiratory system drugs, and cardiovascular drugs. Additionally, information on the physiological targets and therapeutic applications of these drugs was retrieved from the DrugBank database.

2. Biotransformation Prediction

The study utilized a web-based tool called GutBug, which combines artificial intelligence (AI), machine learning, and cheminformatics to predict the biotransformation of biotic and xenobiotic molecules by gut bacterial metabolic enzymes. The GutBug tool was trained on 3,457 enzyme substrates to predict the Enzyme Commission (EC) numbers of gut bacterial enzymes and the gut bacterial strains harboring them. The tool employs a modular design with three modules: the first module predicts the first digit of the EC number (reaction class), the second module predicts the second digit (reaction subclass), and the third module predicts the complete EC number.

3. Database Construction

The study developed a user-friendly web interface for GutBugDB using MySQL, PHP, HTML, and JavaScript. The database includes 363,872 metabolic enzymes from 690 gut bacterial genomes, each annotated with their EC numbers, Expasy IDs, and functional domains. GutBugDB provides information on the gut microbiome enzyme-mediated metabolic biotransformation of 1,439 FDA-approved drugs and nutraceuticals.

4. Data Validation

The study validated GutBugDB using a set of experimentally verified biotic and xenobiotic molecules, including 7 biotic and 10 xenobiotic molecules. The results showed that GutBugDB not only provided information on gut bacteria and enzymes known from the literature but also predicted novel gut bacterial strains and enzymes, further enriching the biotransformation information.

Key Results

1. Database Content

GutBugDB contains 1,439 molecules, including 1,378 FDA-approved drugs and 61 nutraceuticals, classified into 14 pharmacological categories. The database also includes 363,872 metabolic enzymes from 690 gut bacterial strains, each tagged with their EC numbers and functional domains.

2. Prediction Results

GutBugDB provides detailed biotransformation information, including predicted metabolic enzymes and gut bacterial strains capable of metabolizing specific drugs. For example, for the drug L-Dopa, GutBugDB predicted phenol 2-monooxygenase and 6-hydroxy-3-succinoylpyridine 3-monooxygenase from bacteria belonging to the Acinetobacter and Delftia genera, which can metabolize L-Dopa. Additionally, for Flucytosine, GutBugDB predicted cytosine deaminase from bacteria belonging to the Escherichia, Bifidobacterium, and Clostridium genera, which can metabolize Flucytosine.

3. Validation Results

The study validated GutBugDB’s predictions and found that it performed well on known biotransformation information and provided novel gut bacteria and enzyme information. For example, for Lactulose, GutBugDB not only predicted the known phosphorylase and hydrolase enzymes but also predicted that bacteria from the Ruminococcus and Escherichia genera could metabolize Lactulose.

Conclusions and Significance

GutBugDB provides researchers with a comprehensive resource for predicting human gut bacteria-mediated metabolism of drugs and nutraceuticals. The database can help identify potential biotransformation of candidate drug molecules and prevent drug non-responsiveness during prescription, improving drug efficacy and tolerability. Furthermore, GutBugDB offers leads for future experimental validation, advancing the understanding of gut microbiome-mediated biotransformation mechanisms.

Research Highlights

  1. Comprehensiveness: GutBugDB contains 1,439 molecules, including 1,378 FDA-approved drugs and 61 nutraceuticals, providing detailed biotransformation information.
  2. Innovation: The study employed the machine learning-based GutBug tool, which predicts gut bacterial metabolic enzymes and strains, offering higher precision in biotransformation information.
  3. Practicality: GutBugDB serves as an important reference for drug development and prescription, helping researchers better understand the role of the gut microbiota in drug metabolism.

Additional Valuable Information

GutBugDB is scheduled for regular updates, with the most recent update in June 2024. The database is also compatible with new data from future metagenomic studies, allowing it to continuously expand its content and provide ongoing support for gut microbiota research.