GLORYx: Prediction of the Metabolites Resulting from Phase 1 and Phase 2 Biotransformations of Xenobiotics

Abstract

Predicting the structures of metabolites formed in humans can provide advantageous insights for the development of drugs and other compounds. Here we present GLORYx, which integrates machine learning-based site of metabolism (SoM) prediction with reaction rule sets to predict and rank the structures of metabolites that could potentially be formed by phase 1 and/or phase 2 metabolism. GLORYx extends the approach from our previously developed tool GLORY, which predicted metabolite structures for cytochrome P450-mediated metabolism only. A robust approach to ranking the predicted metabolites is attained by using the SoM probabilities predicted by the FAME 3 machine learning models to score the predicted metabolites. On a manually curated test data set containing both phase 1 and phase 2 metabolites, GLORYx achieves a recall of 77% and an area under the receiver operating characteristic curve (AUC) of 0.79. Separate analysis of performance on a large amount of freely available phase 1 and phase 2 metabolite data indicates that achieving a meaningful ranking of predicted metabolites is more difficult for phase 2 than for phase 1 metabolites. GLORYx is freely available as a web server at and is also provided as a software package upon request. The data sets as well as all the reaction rules from this work are also made freely available.

MetaTREE, a Novel Database Focused on Metabolic Trees, Predicts an Important Detoxification Mechanism: The Glutathione Conjugation

“…The ability to predict in silico the metabolism of new chemical entities has attracted great interest in the last years since very common causes of drug failures (such as low efficacy, unsatisfactory pharmacokinetic profile, and toxicity) are often ascribable to an unfavorable impact on drug metabolism [4,5]. Most of the reported predictive studies focus on the redox reactions typically catalyzed by the CYP-450 enzymes [6], while only a few predictive tools for conjugation reactions were reported in the literature [7,8]. This lack of computational studies appears to be especially relevant for both glucuronidations [8,9] and, in particular, reactions with GSH [10] because these metabolic processes are very frequent in drug metabolism and, more importantly, play a key role in the detoxification processes [11].…”

“…Recently, we compiled and manually curated a metabolic database (MetaQSAR) by a meta-analysis of the specialized literature in the 2005-2015 years [17]. This project led to the generation of a database containing about 5000 metabolic reactions, whose reliability has been confirmed by several predictive studies [4,7,8,18]. Notwithstanding this, when used in classification predictive studies, even very accurate metabolic databases show a quite common problem concerning the definition of negative non-substrate molecules.…”

NICEdrug.ch, a workflow for rational drug design and systems-level analysis of drug metabolism

“…To evaluate the NICEdrug.ch reactivity report, we first used an experimental set including 29 small molecules and their 55 unique metabolic products (labeled in public databases) ( Flynn et al, 2020 ). We compared the predictive accuracy of NICEdrug.ch with other tools predicting reactivity, i.e., XenoNet ( Flynn et al, 2020 ), GLORY ( de Bruyn Kops et al, 2021 ; de Bruyn Kops et al, 2019 ), SyGMa ( Ridder and Wagener, 2008 ), and BioTransformer ( Djoumbou-Feunang et al, 2019 ). NICEdrug.ch predicted 53 of the 55 metabolic products from the small molecule dataset, rendering a sensitivity score of 0.96.…”

“…Published frameworks that are comparable to NICEdrug.ch’s reactivity report include GLORY ( de Bruyn Kops et al, 2021 ; de Bruyn Kops et al, 2019 ), BioTransformer ( Djoumbou-Feunang et al, 2019 ), XenoNet ( Flynn et al, 2020 ), SyGMa ( Ridder and Wagener, 2008 ), and other machine-learning based approaches ( Coley et al, 2017 ). All of these methods receive a molecule as an input and predict a set of metabolites as putative products of a metabolic reaction or pathway.…”

Novel advances in biotransformation and bioactivation research—2019 year in review

“…Therefore, inter-departmental collaborative efforts are essential and critical to maximize the highest return on investment from an LSLD strategy. GLORYx: prediction of the metabolites resulting from phase 1 and phase 2 biotransformations of xenobiotics Kops et al (de Bruyn Kops et al 2021) reported a new tool called GLORYx for predicting metabolite structures formed by both Phase 1 and Phase 2 reactions in humans. The approach used the FAst MEtabolizer (FAME) 3 to predict the likelihood for metabolism at every atom in a molecule using extremely randomized trees.…”