GutBug: A tool for prediction of human gut bacteria mediated biotransformation of biotic and xenobiotic molecules using machine learning

Journal home page for Journal of Molecular BiologyAuthor links open overlay panel, , Abstract

Dietary components and bioactive molecules present in functional foods and nutraceuticals provide various beneficial effects including modulation of host gut microbiome. These metabolites along with orally administered drugs can be potentially bio-transformed by gut microbiome, which can alter their bioavailability and intended biological or pharmacological activity resulting in individual or population-specific variation in drug and dietary responses. Experimental determination of microbiome-mediated metabolism of orally ingested molecules is difficult due to the enormous diversity and complexity of the gut microbiome. To address this problem, we developed “GutBug”, a web-based resource that predicts all possible bacterial metabolic enzymes that can potentially biotransform xenobiotics and biotic molecules using a combination of machine learning, neural networks and chemoinformatic methods. Using 3,457 enzyme substrates for training and a curated database of 363,872 enzymes from ∼700 gut bacterial strains, GutBug can predict complete EC number of the bacterial enzymes involved in a biotransformation reaction of the given molecule along with the reaction centres with accuracies between 0.78 to 0.97 across different reaction classes. Validation of GutBug’s performance using 27 molecules known to be biotransformed by human gut bacteria, including complex polysaccharides, flavonoids, and oral drugs further attests to GutBug’s accuracy and utility. Thus, GutBug enhances our understanding of various metabolite-gut bacterial interactions and their resultant effects on the human host health across populations, which will find enormous applications in diet design and intervention, identification and administration of new prebiotics, development of nutraceutical products, and improvements in drug designing. GutBug is available at https://metabiosys.iiserb.ac.in/gutbug.

Section snippetsIntroduction:

The human gut microbiome (HGM) constitutes more than 3.3 million unique genes from more than thousand different human gut bacterial species, and provides additional metabolic activity to the host to metabolise molecules that escape host-mediated digestion and metabolism [1]. HGM metabolizes an ingested food to produce an extensive range of microbial metabolites that play a crucial role in human health and physiology, which makes it a potential target to modulate health and disease [2].

Gut bacterial genomes, metabolic enzymes, and substrates databases

For developing a machine learning based tool to predict metabolic enzymes from gut bacteria capable of metabolizing biotic or xenobiotic molecules, we constructed an extensive database of all metabolic enzymes and their substrates present in all known human gut bacteria. The list and date of all available human gut bacteria was retrieved from various databases like HMP (Human Microbiome Project) [13], VMH (Virtual Metabolic Human) [14] and DrugBug [12]. A total of 457 strains were obtained from

Calculation of features for substrate molecules

For training machine learning models, different types of features encoding structural and chemical information were calculated from the training set of molecules. Molecular descriptors were calculated using RDKit (RDKit: Open-source cheminformatics; http://www.rdkit.org) to get 196 features that represent chemical as well as physical properties of molecules. Structural information in form of Linear Fingerprints and Circular Fingerprints was also calculated. Circular fingerprints generated using

Metabolic complexity of the constructed database

Distribution of substrates into the six reaction classes was observed to be highly imbalanced. Oxidoreductase (EC1) class of enzymes had the highest number of substrates whereas ligase (EC6) and isomerase (EC5) classes had the least number of substrates participating in their reactions (Figure S1a, Table S3a). Number of substrates participating in a single type of reaction (non-redundant EC substrates) were also higher compared to the number of substrates participating in multiple reactions

Module-1: Prediction of reaction class

Prediction of reaction class (first digit of EC number) of any enzyme metabolizing a molecule is a multiclass multilabel problem as multiple reaction classes exist (multiclass) and a molecule can be metabolised by enzymes from different classes (multilabel). Further, since substrate dataset is highly disproportionate across the six reaction classes, different types of approaches were undertaken to develop a classifier for the first module.

The binary classification approach was adopted since

Module-2: Prediction of reaction subclass

Prediction of reaction subclass (second digit of EC number) is a multilabel problem as within a particular reaction class multiple subclasses exist with varying number of molecules present in these subclasses. Moreover, since each reaction class has different number of subclasses (Table S5), prediction of reaction subclass for each EC class was carried out individually, and thus six different multilabel models were developed. Further, the number of training molecules for prediction of subclass

Module-3: Prediction of complete EC number of enzymes and identification of reaction centres

In the final module, complete EC number of enzymes that can metabolise a given molecule was determined using molecular similarity search using Tanimoto coefficient, which is a similarity index that calculates structural and elemental overlap between two molecules. Higher the Tanimoto index, higher is the similarity between two molecules. The Tanimoto threshold value as mentioned in Table 2 of Results refers to the threshold at which an input molecule was found similar to the molecules in the

Development of GutBug web server

GutBug is deployed as a user-friendly webserver available at https://metabiosys.iiserb.ac.in/gutbug. Users can upload a sdf file or input the PubChem ID of the molecule. A single step pipeline integrating all the modules provides result in the form of five tables. Summary of the predictions is provided in the first table R1 that displays the predicted EC number, name of the enzyme, and the bacterial strains that harbour those enzymes. Table R2 displays information about EC numbers at the

Performance of GutBug on validation set

Many studies have experimentally identified orally administered molecules that are metabolised by human gut bacteria. However, in many cases, either the metabolizing enzyme or the bacteria harbouring the enzymes are still unidentified [21]. The performance of GutBug was examined using a validation set that included a diverse array of dietary substrates like quercetin, curcumin (flavonoids), chalcone, inulin (polysaccharide), lactulose (synthetic carbohydrate), 2’-fucosyllactose and

Discussion

HGM provides additional pool of metabolic enzymes to human host to metabolise dietary components and affect the bioavailability of bioactive molecules in food/nutraceutical products that in turn affects the overall composition and health of the gut microbiome 2, 38. Microbiome derived metabolism also has pharmaceutical significance since the promiscuous metabolism of drugs by gut bacteria can modify the potency/toxicity of such drugs thus altering their pharmacokinetic and pharmacodynamic

Materials and methods:

Details about all the methods used for development of GutBug are available in Supplementary methods section.

Acknowledgements:

ASM and GNS thank IISER Bhopal for fellowship. This work was supported by the financial grant (BT/PR34239/AI/133/23/2019) by the Department of Biotechnology, Government of India.

References: (44)C.M. Guinane et al.Role of the gut microbiota in health and chronic gastrointestinal disease: understanding a hidden metabolic organ

Therap. Adv. Gastroenterol.

(Jul. 2013)

C. L. Gentile and T. L. Weir, “The gut microbiota at the intersection of diet and human health,” Science (80-. )., vol....A. Cencic et al.The Role of Functional Foods, Nutraceuticals, and Food Supplements in Intestinal Health

Nutrients

(Jun. 2010)

E.F. Enright et al.The Impact of the Gut Microbiota on Drug Metabolism and Clinical Outcome

Yale J. Biol. Med.

(2016)

E. OrgGenetic and environmental control of host-gut microbiota interactions

Genome Res.

(Oct. 2015)

V. Prasoodanan P. K. et al., “Western and non-western gut microbiomes reveal new roles of Prevotella in carbohydrate...Y.K. TamIndividual Variation in First-Pass Metabolism

Clin. Pharmacokinet.

(Oct. 1993)

G. CrucianiMetaSite: Understanding Metabolism in Human Cytochromes from the Perspective of the Chemist

J. Med. Chem.

(Nov. 2005)

J. KirchmairFAst MEtabolizer (FAME): A Rapid and Accurate Predictor of Sites of Metabolism in Multiple Species by Endogenous Enzymes

J. Chem. Inf. Model.

(Nov. 2013)

F. ChengadmetSAR: A Comprehensive Source and Free Tool for Assessment of Chemical ADMET Properties

J. Chem. Inf. Model.

(Nov. 2012)

A.K. Sharma et al.A novel approach for the prediction of species-specific biotransformation of xenobiotic/drug molecules by the human gut microbiota

Sci. Rep.

(Dec. 2017)

P.J. Turnbaugh et al.The Human Microbiome Project

Nature

(Oct. 2007)

View full text

© 2023 Elsevier Ltd. All rights reserved.

留言 (0)

沒有登入
gif