Knowledge-based expert systems for toxicity and metabolism prediction: DEREK, StAR and METEOR

Identification of ligands binding to MB327-PAM-1, a binding pocket relevant for resensitization of nAChRs

Desensitization of nicotinic acetylcholine receptors (nAChRs) can be induced by overstimulation with acetylcholine (ACh) caused by an insufficient degradation of ACh after poisoning with organophosphorus compounds (OPCs). Currently, there is no generally applicable treatment for OPC poisoning that directly targets the desensitized nAChR. The bispyridinium compound MB327, an allosteric modulator of nAChR, has been shown to act as a resensitizer of nAChRs, indicating that drugs binding directly to nAChRs can have beneficial effects after OPC poisoning. However, MB327 also acts as an inhibitor of nAChRs at higher concentrations and can thus not be used for OPC poisoning treatment. Consequently, novel, more potent resensitizers are required. To successfully design novel ligands, the knowledge of the binding site is of utmost importance. Recently, we performed in silico studies to identify a new potential binding site of MB327, MB327-PAM-1, for which a more affine ligand, UNC0646, has been described. In this work, we performed ligand-based screening approaches to identify novel analogs of UNC0646 to help further understand the structure-affinity relationship of this compound class. Furthermore, we used structure-based screenings and identified compounds representing four new chemotypes binding to MB327-PAM-1. One of these compounds, cycloguanil, is the active metabolite of the antimalaria drug proguanil and shows a higher affinity towards MB327-PAM-1 than MB327. Furthermore, cycloguanil can reestablish the muscle force in soman-inhibited rat muscles. These results can act as a starting point to develop more potent resensitizers of nAChR and to close the gap in the treatment after OPC poisoning.

Should Transformation Products Change the Way We Manage Chemicals?

When chemical pollutants enter the environment, they can undergo diverse transformation processes, forming a wide range of transformation products (TPs), some of them benign and others more harmful than their precursors. To date, the majority of TPs remain largely unrecognized and unregulated, particularly as TPs are generally not part of routine chemical risk or hazard assessment. Since many TPs formed from oxidative processes are more polar than their precursors, they may be especially relevant in the context of persistent, mobile, and toxic (PMT) and very persistent and very mobile (vPvM) substances, which are two new hazard classes that have recently been established on a European level. We highlight herein that as a result, TPs deserve more attention in research, chemicals regulation, and chemicals management. This perspective summarizes the main challenges preventing a better integration of TPs in these areas: (1) the lack of reliable high-throughput TP identification methods, (2) uncertainties in TP prediction, (3) inadequately considered TP formation during (advanced) water treatment, and (4) insufficient integration and harmonization of TPs in most regulatory frameworks. A way forward to tackle these challenges and integrate TPs into chemical management is proposed.

DeepAmes: A deep learning-powered Ames test predictive model with potential for regulatory application

The Ames assay is required by the regulatory agencies worldwide to assess the mutagenic potential risk of consumer products. As well as this in vitro assay, in silico approaches have been widely used to predict Ames test results as outlined in the International Council for Harmonization (ICH) guidelines. Building on this in silico approach, here we describe DeepAmes, a high performance and robust model developed with a novel deep learning (DL) approach for potential utility in regulatory science. DeepAmes was developed with a large and consistent Ames dataset (>10,000 compounds) and was compared with other five standard Machine Learning (ML) methods. Using a test set of 1,543 compounds, DeepAmes was the best performer in predicting the outcome of Ames assay. In addition, DeepAmes yielded the best and most stable performance up to when compounds were >30% outside of the applicability domain (AD). Regarding the potential for regulatory application, a revised version of DeepAmes with a much-improved sensitivity of 0.87 from 0.47. In conclusion, DeepAmes provides a DL-powered Ames test predictive model for predicting the results of Ames tests; with its defined AD and clear context of use, DeepAmes has potential for utility in regulatory application.

Conducting bioinformatics analysis to predict sulforaphane-triggered adverse outcome pathways in healthy human cells

Sulforaphane (SFN) is a naturally occurring molecule present in plants from Brassica family. It becomes bioactive after hydrolytic reaction mediated by myrosinase or human gastrointestinal microbiota. Sulforaphane gained scientific popularity due to its antioxidant and anti-cancer properties. However, its toxicity profile and potential to cause adverse effects remain largely unidentified. Thus, this study aimed to generate SFN-triggered adverse outcome pathway (AOP) by looking at the relationship between SFN-chemical structure and its toxicity, as well as SFN-gene interactions. Quantitative structure-activity relationship (QSAR) analysis identified 2 toxophores (Derek Nexus software) that have the potential to cause chromosomal damage and skin sensitization in mammals or mutagenicity in bacteria. Data extracted from Comparative Toxicogenomics Database (CTD) linked SFN with previously proposed outcomes via gene interactions. The total of 11 and 146 genes connected SFN with chromosomal damage and skin diseases, respectively. However, network analysis (NetworkAnalyst tool) revealed that these genes function in wider networks containing 490 and 1986 nodes, respectively. The over-representation analysis (ExpressAnalyst tool) pointed out crucial biological pathways regulated by SFN-interfering genes. These pathways are uploaded to AOP-helpFinder tool which found the 2321 connections between 19 enriched pathways and SFN which were further considered as key events. Two major, interconnected AOPs were generated: first starting from disruption of biological pathways involved in cell cycle and cell proliferation leading to increased apoptosis, and the second one connecting activated immune system signaling pathways to inflammation and apoptosis. In both cases, chromosomal damage and/or skin diseases such as dermatitis or psoriasis appear as adverse outcomes.

Ligand-based virtual screening, molecular dynamics, and biological evaluation of repurposed drugs as inhibitors of Trypanosoma cruzi proteasome

Chagas disease is a well-known Neglected Tropical Disease, mostly endemic in continental Latin America, but that has spread to North America and Europe. Unfortunately, current treatments against such disease are ineffective and produce known and undesirable side effects. To find novel effective drug candidates to treat Chagas disease, we uniquely explore the Trypanosoma cruzi proteasome as a recent biological target and, also, apply drug repurposing through different computational methodologies. For this, we initially applied protein homology modeling to build a robust model of proteasome β4/β5 subunits, since there is no crystallographic structure of this target. Then, we used it on a drug repurposing via a virtual screening campaign starting with more than 8,000 drugs and including the methodologies: ligand-based similarity, toxicity predictions, and molecular docking. Three drugs were selected concerning their favorable interactions at the protein binding site and subsequently submitted to molecular dynamics simulations, which allowed us to elucidate their behavior and compare such theoretical results with experimental ones, obtained in biological assays also described in this paper.Communicated by Ramaswamy H. Sarma.

Genotoxicity evaluation of a valsartan-related complex N-nitroso-impurity

Recently, the formation of genotoxic and carcinogenic N-nitrosamines impurities during drug manufacturing of tetrazole-containing angiotensin-II blockers has been described. However, drug-related (complex) nitrosamines may also be generated under certain conditions, i.e., through nitrosation of vulnerable amines in drug substances in the presence of nitrite. An investigation of valsartan drug substance showed that a complex API-related N-nitrosamine chemically designated as (S)-2-(((2'-(1H-tetrazol-5-yl)-[1,1'-biphenyl]-4-yl)methyl)(nitroso)amino)-3-methylbutanoic acid (named 181-14) may be generated. 181-14 was shown to be devoid of a mutagenic potential in the Non-GLP Ames test. According to ICH M7 (R1) (2018), impurities that are not mutagenic in the Ames test would be considered Class 5 impurities and limited according to ICH Q3A (R2) and B (R2) (2006) guidelines. However, certain regulatory authorities raised the concern that the Ames test may not be sufficiently sensitive to detect a mutagenic potential of nitrosamines and requested a confirmatory in vivo study using a transgenic animal genotoxicity model. Our data show that 181-14 was not mutagenic in the transgenic gene mutation assay in Muta^TMMice. The data support the conclusion that the Ames test is an adequate and sensitive test system to assess a mutagenic potential of nitrosamines.

In-Silico Drug Toxicity and Interaction Prediction for Plant Complexes Based on Virtual Screening and Text Mining

Potential drug toxicities and drug interactions of redundant compounds of plant complexes may cause unexpected clinical responses or even severe adverse events. On the other hand, super-additivity of drug interactions between natural products and synthetic drugs may be utilized to gain better performance in disease management. Although without enough datasets for prediction model training, based on the SwissSimilarity and PubChem platforms, for the first time, a feasible workflow of prediction of both toxicity and drug interaction of plant complexes was built in this study. The optimal similarity score threshold for toxicity prediction of this system is 0.6171, based on an analysis of 20 different herbal medicines. From the PubChem database, 31 different sections of toxicity information such as "Acute Effects", "NIOSH Toxicity Data", "Interactions", "Hepatotoxicity", "Carcinogenicity", "Symptoms", and "Human Toxicity Values" sections have been retrieved, with dozens of active compounds predicted to exert potential toxicities. In Spatholobus suberectus Dunn (SSD), there are 9 out of 24 active compounds predicted to play synergistic effects on cancer management with various drugs or factors. The synergism between SSD, luteolin and docetaxel in the management of triple-negative breast cancer was proved by the combination index assay, synergy score detection assay, and xenograft model.

Biophysical and pharmacokinetic characterization of a small-molecule inhibitor of RUNX1/ETO tetramerization with anti-leukemic effects

Acute myeloid leukemia (AML) is a malignant disease of immature myeloid cells and the most prevalent acute leukemia among adults. The oncogenic homo-tetrameric fusion protein RUNX1/ETO results from the chromosomal translocation t(8;21) and is found in AML patients. The nervy homology region 2 (NHR2) domain of ETO mediates tetramerization; this oligomerization is essential for oncogenic activity. Previously, we identified the first-in-class small-molecule inhibitor of NHR2 tetramer formation, 7.44, which was shown to specifically interfere with NHR2, restore gene expression down-regulated by RUNX1/ETO, inhibit the proliferation of RUNX1/ETO-depending SKNO-1 cells, and reduce the RUNX1/ETO-related tumor growth in a mouse model. However, no biophysical and structural characterization of 7.44 binding to the NHR2 domain has been reported. Likewise, the compound has not been characterized as to physicochemical, pharmacokinetic, and toxicological properties. Here, we characterize the interaction between the NHR2 domain of RUNX1/ETO and 7.44 by biophysical assays and show that 7.44 interferes with NHR2 tetramer stability and leads to an increase in the dimer population of NHR2. The affinity of 7.44 with respect to binding to NHR2 is K_lig = 3.75 ± 1.22 µM. By NMR spectroscopy combined with molecular dynamics simulations, we show that 7.44 binds with both heteroaromatic moieties to NHR2 and interacts with or leads to conformational changes in the N-termini of the NHR2 tetramer. Finally, we demonstrate that 7.44 has favorable physicochemical, pharmacokinetic, and toxicological properties. Together with biochemical, cellular, and in vivo assessments, the results reveal 7.44 as a lead for further optimization towards targeted therapy of t(8;21) AML.

A Computational Approach Applied to the Study of Potential Allosteric Inhibitors Protease NS2B/NS3 from Dengue Virus

Dengue virus (DENV) is a danger to more than 400 million people in the world, and there is no specific treatment. Thus, there is an urgent need to develop an effective method to combat this pathology. NS2B/NS3 protease is an important biological target due it being necessary for viral replication and the fact that it promotes the spread of the infection. Thus, this study aimed to design DENV NS2B/NS3pro allosteric inhibitors from a matrix compound. The search was conducted using the Swiss Similarity tool. The compounds were subjected to molecular docking calculations, molecular dynamics simulations (MD) and free energy calculations. The molecular docking results showed that two compounds, ZINC000001680989 and ZINC000001679427, were promising and performed important hydrogen interactions with the Asn152, Leu149 and Ala164 residues, showing the same interactions obtained in the literature. In the MD, the results indicated that five residues, Lys74, Leu76, Asn152, Leu149 and Ala166, contribute to the stability of the ligand at the allosteric site for all of the simulated systems. Hydrophobic, electrostatic and van der Waals interactions had significant effects on binding affinity. Physicochemical properties, lipophilicity, water solubility, pharmacokinetics, druglikeness and medicinal chemistry were evaluated for four compounds that were more promising, showed negative indices for the potential penetration of the Blood Brain Barrier and expressed high human intestinal absorption, indicating a low risk of central nervous system depression or drowsiness as the the side effects. The compound ZINC000006694490 exhibited an alert with a plausible level of toxicity for the purine base chemical moiety, indicating hepatotoxicity and chromosome damage in vivo in mouse, rat and human organisms. All of the compounds selected in this study showed a synthetic accessibility (SA) score lower than 4, suggesting the ease of new syntheses. The results corroborate with other studies in the literature, and the computational approach used here can contribute to the discovery of new and potent anti-dengue agents.

BioTransformer 3.0-a web server for accurately predicting metabolic transformation products

BioTransformer 3.0 (https://biotransformer.ca) is a freely available web server that supports accurate, rapid and comprehensive in silico metabolism prediction. It combines machine learning approaches with a rule-based system to predict small-molecule metabolism in human tissues, the human gut as well as the external environment (soil and water microbiota). Simply stated, BioTransformer takes a molecular structure as input (SMILES or SDF) and outputs an interactively sortable table of the predicted metabolites or transformation products (SMILES, PNG images) along with the enzymes that are predicted to be responsible for those reactions and richly annotated downloadable files (CSV and JSON). The entire process typically takes less than a minute. Previous versions of BioTransformer focused exclusively on predicting the metabolism of xenobiotics (such as plant natural products, drugs, cosmetics and other synthetic compounds) using a limited number of pre-defined steps and somewhat limited rule-based methods. BioTransformer 3.0 uses much more sophisticated methods and incorporates new databases, new constraints and new prediction modules to not only more accurately predict the metabolic transformation products of exogenous xenobiotics but also the transformation products of endogenous metabolites, such as amino acids, peptides, carbohydrates, organic acids, and lipids. BioTransformer 3.0 can also support customized sequential combinations of these transformations along with multiple iterations to simulate multi-step human biotransformation events. Performance tests indicate that BioTransformer 3.0 is 40–50% more accurate, far less prone to combinatorial ‘explosions’ and much more comprehensive in terms of metabolite coverage/capabilities than previous versions of BioTransformer.

Increasing the Value of Data Within a Large Pharmaceutical Company Through In Silico Models

The present contribution describes how in silico models and methods are applied at different stages of the drug discovery process in the pharmaceutical industry. A description of the most relevant computational methods and tools is given along with an evaluation of their performance in the assessment of potential genotoxic impurities and the prediction of off-target in vitro pharmacology. The challenges of predicting the outcome of highly complex in vivo studies are discussed followed by considerations on how novel ways to manage, store, exchange, and analyze data may advance knowledge and facilitate modeling efforts. In this context, the current status of broad data sharing initiatives, namely, eTOX and eTransafe, will be described along with related projects that could significantly reduce the use of animals in drug discovery in the future.

In Silico Models for Predicting Acute Systemic Toxicity

In this chapter, we give a brief overview of the regulatory requirements for acute systemic toxicity information in the European Union, and we review structure-based computational models that are available and potentially useful in the assessment of acute systemic toxicity. Emphasis is placed on quantitative structure-activity relationship (QSAR) models implemented by means of a range of software tools. The most recently published literature models for acute systemic toxicity are also discussed, and perspectives for future developments in this field are offered.

Natural Products-Based Drug Design against SARS-CoV-2 Mpro 3CLpro

Coronavirus disease 2019 (COVID-19), caused by severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2), has received global attention due to the serious threat it poses to public health. Since the outbreak in December 2019, millions of people have been affected and its rapid global spread has led to an upsurge in the search for treatment. To discover hit compounds that can be used alone or in combination with repositioned drugs, we first analyzed the pharmacokinetic and toxicological properties of natural products from Brazil's semiarid region. After, we analyzed the site prediction and druggability of the SARS-CoV-2 main protease (Mpro), followed by docking and molecular dynamics simulation. The best SARS-CoV-2 Mpro complexes revealed that other sites were accessed, confirming that our approach could be employed as a suitable starting protocol for ligand prioritization, reinforcing the importance of catalytic cysteine-histidine residues and providing new structural data that could increase the antiviral development mainly against SARS-CoV-2. Here, we selected 10 molecules that could be in vitro assayed in response to COVID-19. Two compounds (b01 and b02) suggest a better potential for interaction with SARS-CoV-2 Mpro and could be further studied.

MetaClass, a Comprehensive Classification System for Predicting the Occurrence of Metabolic Reactions Based on the MetaQSAR Database

(1) Background: Machine learning algorithms are finding fruitful applications in predicting the ADME profile of new molecules, with a particular focus on metabolism predictions. However, the development of comprehensive metabolism predictors is hampered by the lack of highly accurate metabolic resources. Hence, we recently proposed a manually curated metabolic database (MetaQSAR), the level of accuracy of which is well suited to the development of predictive models. (2) Methods: MetaQSAR was used to extract datasets to predict the metabolic reactions subdivided into major classes, classes and subclasses. The collected datasets comprised a total of 3788 first-generation metabolic reactions. Predictive models were developed by using standard random forest algorithms and sets of physicochemical, stereo-electronic and constitutional descriptors. (3) Results: The developed models showed satisfactory performance, especially for hydrolyses and conjugations, while redox reactions were predicted with greater difficulty, which was reasonable as they depend on many complex features that are not properly encoded by the included descriptors. (4) Conclusions: The generated models allowed a precise comparison of the propensity of each metabolic reaction to be predicted and the factors affecting their predictability were discussed in detail. Overall, the study led to the development of a freely downloadable global predictor, MetaClass, which correctly predicts 80% of the reported reactions, as assessed by an explorative validation analysis on an external dataset, with an overall MCC = 0.44.

Holistic evaluation of biodegradation pathway prediction: assessing multi-step reactions and intermediate products

The prediction of metabolism and biotransformation pathways of xenobiotics is a highly desired tool in environmental sciences, drug discovery, and (eco)toxicology. Several systems predict single transformation steps or complete pathways as series of parallel and subsequent steps. Their performance is commonly evaluated on the level of a single transformation step. Such an approach cannot account for some specific challenges that are caused by specific properties of biotransformation experiments. That is, missing transformation products in the reference data that occur only in low concentrations, e.g. transient intermediates or higher-generation metabolites. Furthermore, some rule-based prediction systems evaluate the performance only based on the defined set of transformation rules. Therefore, the performance of these models cannot be directly compared. In this paper, we introduce a new evaluation framework that extends the evaluation of biotransformation prediction from single transformations to whole pathways, taking into account multiple generations of metabolites. We introduce a procedure to address transient intermediates and propose a weighted scoring system that acknowledges the uncertainty of higher-generation metabolites. We implemented this framework in enviPath and demonstrate its strict performance metrics on predictions of in vitro biotransformation and degradation of xenobiotics in soil. Our approach is model-agnostic and can be transferred to other prediction systems. It is also capable of revealing knowledge gaps in terms of incompletely defined sets of transformation rules.

Machine learning approaches for elucidating the biological effects of natural products

Covering: 2000 to 2020 Machine learning (ML) is an efficient tool for the prediction of bioactivity and the study of structure-activity relationships. Over the past decade, an emerging trend for combining these approaches with the study of natural products (NPs) has developed in order to manage the challenge of the discovery of bioactive NPs. In the present review, we will introduce the basic principles and protocols for using the ML approach to investigate the bioactivity of NPs, citing a series of practical examples regarding the study of anti-microbial, anti-cancer, and anti-inflammatory NPs, etc. ML algorithms manage a variety of classification and regression problems associated with bioactive NPs, from those that are linear to non-linear and from pure compounds to plant extracts. Inspired by cases reported in the literature and our own experience, a number of key points have been emphasized for reducing modeling errors, including dataset preparation and applicability domain analysis.

Machine Learning in Predictive Toxicology: Recent Applications and Future Directions for Classification Models

In recent times, machine learning has become increasingly prominent in predictive toxicology as it has shifted from in vivo studies toward in silico studies. Currently, in vitro methods together with other computational methods such as quantitative structure-activity relationship modeling and absorption, distribution, metabolism, and excretion calculations are being used. An overview of machine learning and its applications in predictive toxicology is presented here, including support vector machines (SVMs), random forest (RF) and decision trees (DTs), neural networks, regression models, naïve Bayes, k-nearest neighbors, and ensemble learning. The recent successes of these machine learning methods in predictive toxicology are summarized, and a comparison of some models used in predictive toxicology is presented. In predictive toxicology, SVMs, RF, and DTs are the dominant machine learning methods due to the characteristics of the data available. Lastly, this review describes the current challenges facing the use of machine learning in predictive toxicology and offers insights into the possible areas of improvement in the field.

Seeing the forest for the trees: Retrieving plant secondary biochemical pathways from metabolome networks

Over the last decade, a giant leap forward has been made in resolving the main bottleneck in metabolomics, i.e., the structural characterization of the many unknowns. This has led to the next challenge in this research field: retrieving biochemical pathway information from the various types of networks that can be constructed from metabolome data. Searching putative biochemical pathways, referred to as biotransformation paths, is complicated because several flaws occur during the construction of metabolome networks. Multiple network analysis tools have been developed to deal with these flaws, while in silico retrosynthesis is appearing as an alternative approach. In this review, the different types of metabolome networks, their flaws, and the various tools to trace these biotransformation paths are discussed.

Machine Learning Predicts Degree of Aromaticity from Structural Fingerprints

Prediction of whether a compound is "aromatic" is at first glance a relatively simple task-does it obey Hückel's rule (planar cyclic π-system with 4n + 2 electrons) or not? However, aromaticity is far from a binary property, and there are distinct variations in the chemical and biological behavior of different systems which obey Hückel's rule and are thus classified as aromatic. To that end, the aromaticity of each molecule in a large public dataset was quantified by an extension of the work of Raczyńska et al. Building on this data, a method is proposed for machine learning the degree of aromaticity of each aromatic ring in a molecule. Categories are derived from the numeric results, allowing the differentiation of structural patterns between them and thus a better representation of the underlying chemical and biological behavior in expert and (Q)SAR systems.

Virtual screening, ADME/Tox predictions and the drug repurposing concept for future use of old drugs against the COVID-19

The new Coronavirus (SARS-CoV-2) is the cause of a serious infection in the respiratory tract called COVID-19. Structures of the main protease of SARS-CoV-2 (M^pro), responsible for the replication of the virus, have been solved and quickly made available, thus allowing the design of compounds that could interact with this protease and thus to prevent the progression of the disease by avoiding the viral peptide to be cleaved, so that smaller viral proteins can be released into the host's plasma. These structural data are extremely important for in silico design and development of compounds as well, being possible to quick and effectively identify potential inhibitors addressed to such enzyme's structure. Therefore, in order to identify potential inhibitors for M^pro, we used virtual screening approaches based with the structure of the enzyme and two compounds libraries, targeted to SARS-CoV-2, containing compounds with predicted activity against M^pro. In this way, we selected, through docking studies, the 100 top-ranked compounds, which followed to subsequent studies of pharmacokinetic and toxicity predictions. After all the simulations and predictions here performed, we obtained 10 top-ranked compounds that were again in silico analyzed inside the M^pro catalytic site, together some drugs that are being currently investigated for treatment of COVID-19. After proposing and analyzing the interaction modes of these compounds, we submitted one molecule then selected as template to a 2D similarity study in a database containing drugs approved by FDA and we have found and indicated Apixaban as a potential drug for future treatment of COVID-19.

•

The new coronavirus (SARS-CoV-2) is the cause of a serious infection in the respiratory tract called COVID-19.

•

The main protease SARS-CoV-2 (M^pro) is essential in the process of maturation and infectivity of the virus.

•

In silico methodologies are extremely important to identify potential inhibitors for the target structure quickly and effectively.

•

The drug repurposing is an important concept for future use of old drugs.

Dual graph convolutional neural network for predicting chemical networks

Background

Predicting of chemical compounds is one of the fundamental tasks in bioinformatics and chemoinformatics, because it contributes to various applications in metabolic engineering and drug discovery. The recent rapid growth of the amount of available data has enabled applications of computational approaches such as statistical modeling and machine learning method. Both a set of chemical interactions and chemical compound structures are represented as graphs, and various graph-based approaches including graph convolutional neural networks have been successfully applied to chemical network prediction. However, there was no efficient method that can consider the two different types of graphs in an end-to-end manner.

Results

We give a new formulation of the chemical network prediction problem as a link prediction problem in a graph of graphs (GoG) which can represent the hierarchical structure consisting of compound graphs and an inter-compound graph. We propose a new graph convolutional neural network architecture called dual graph convolutional network that learns compound representations from both the compound graphs and the inter-compound network in an end-to-end manner.

Conclusions

Experiments using four chemical networks with different sparsity levels and degree distributions shows that our dual graph convolution approach achieves high prediction performance in relatively dense networks, while the performance becomes inferior on extremely-sparse networks.

KnowTox: pipeline and case study for confident prediction of potential toxic effects of compounds in early phases of development

Risk assessment of newly synthesised chemicals is a prerequisite for regulatory approval. In this context, in silico methods have great potential to reduce time, cost, and ultimately animal testing as they make use of the ever-growing amount of available toxicity data. Here, KnowTox is presented, a novel pipeline that combines three different in silico toxicology approaches to allow for confident prediction of potentially toxic effects of query compounds, i.e. machine learning models for 88 endpoints, alerts for 919 toxic substructures, and computational support for read-across. It is mainly based on the ToxCast dataset, containing after preprocessing a sparse matrix of 7912 compounds tested against 985 endpoints. When applying machine learning models, applicability and reliability of predictions for new chemicals are of utmost importance. Therefore, first, the conformal prediction technique was deployed, comprising an additional calibration step and per definition creating internally valid predictors at a given significance level. Second, to further improve validity and information efficiency, two adaptations are suggested, exemplified at the androgen receptor antagonism endpoint. An absolute increase in validity of 23% on the in-house dataset of 534 compounds could be achieved by introducing KNNRegressor normalisation. This increase in validity comes at the cost of efficiency, which could again be improved by 20% for the initial ToxCast model by balancing the dataset during model training. Finally, the value of the developed pipeline for risk assessment is discussed using two in-house triazole molecules. Compared to a single toxicity prediction method, complementing the outputs of different approaches can have a higher impact on guiding toxicity testing and de-selecting most likely harmful development-candidate compounds early in the development process.

New in silico models to predict in vitro micronucleus induction as marker of genotoxicity

The evaluation of genotoxicity is a fundamental part of the safety assessment of chemicals due to the relevance of the potential health effects of genotoxicants. Among the testing methods available, the in vitro micronucleus assay with mammalian cells is one of the most used and required by regulations targeting several industrial sectors such as the cosmetic industry and food-related sectors. As an alternative to the testing methods, in recent years, lots in silico methods were developed to predict the genotoxicity of chemicals, including models for the Ames mutagenicity test, the in vitro chromosomal aberrations and the in vivo micronucleus assay. We developed several in silico models for the prediction of genotoxicity as reflected by the in vitro micronucleus assay. The resulting models include both statistical and knowledge-based models. The most promising model is the one based on fragments extracted with the SARpy platform. More than 100 structural alerts were extracted, including also fragments associated with the non-genotoxic activity. The model is characterized by high accuracy and the lowest false negative rate, making this tool suitable for chemical screening according to the regulators' needs. The SARpy model will be implemented on the VEGA platform (https://www.vegahub.eu) and will be freely available.

Search for Potential Inducible Nitric Oxide Synthase Inhibitors with Favorable ADMET Profiles for the Therapy of Helicobacter pylori Infections

Background:

Helicobacter pylori is a gram-negative bacterium related to chronic gastritis, peptic ulcer and gastric carcinoma. During its infection process, promotes excessive inflammatory response, increasing the release of reactive species and inducing the production of pro-inflammatory mediators. Inducible Nitric Oxide Synthase (iNOS) plays a crucial role in the gastric carcinogenesis process and a key mediator of inflammation and host defense systems, which is expressed in macrophages induced by inflammatory stimuli. In chronic diseases such as Helicobacter pylori infections, the overproduction of NO due to the prolonged induction of iNOS is of major concern.

Objective:

In this sense, the search for potential iNOS inhibitors is a valuable strategy in the overall process of Helicobacter pylori pathogeny.

Method:

In silico techniques were applied in the search of interesting compounds against Inducible Nitric Oxide Synthase enzyme in a chemical space of natural products and derivatives from the Analyticon Discovery databases.

Results:

The five compounds with the best iNOS inhibition profile were selected for activity and toxicity predictions. Compound 9 (CAS 88198-99-6) displayed significant potential for iNOS inhibition, forming hydrogen bonds with residues from the active site and an ionic interaction with heme. This compound also displayed good bioavailability and absence of toxicity/or from its probable metabolites.

Conclusion:

The top-ranked compounds from the virtual screening workflow show promising results regarding the iNOS inhibition profile. The results evidenced the importance of the ionic bonding during docking selection, playing a crucial role in binding and positioning during ligand-target selection for iNOS.

“Drug-likeness” properties of natural compounds

Our previous work was focused on the fundamental physical and chemical concepts behind “drug-likeness” and “natural product (NP)-likeness”. Herein, we discuss further details on the concepts of “drug-likeness”, “lead-likeness” and “NP-likeness”. The discussion will first focus on NPs as drugs, then a discussion of previous studies in which the complexities of the scaffolds and chemical space of naturally occurring compounds have been compared with synthetic, semisynthetic compounds and the Food and Drug Administration-approved drugs. This is followed by guiding principles for designing “drug-like” natural product libraries for lead compound discovery purposes. In addition, we present a tool for measuring “NP-likeness” of compounds and a brief presentation of machine-learning approaches. A binary quantitative structure–activity relationship for classifying drugs from nondrugs and natural compounds from nonnatural ones is also described. While the studies add to the plethora of recently published works on the “drug-likeness” of NPs, it no doubt increases our understanding of the physicochemical properties that make NPs fall within the ranges associated with “drug-like” molecules.

Toxicity profiling and prioritization of plant-derived antimalarial agents

Human malaria is the most widespread mosquito-borne life-threatening disease worldwide. In the absence of effective vaccines, prevention and treatment of malaria only depend on prophylaxis and drug-based therapy either in monotherapy or in combination. Unfortunately, the number of available antimalarial drugs presenting different mechanisms of action is rather limited. In addition, the appearance of drug-resistance in the parasite strains impacts the efficacy of the treatments. As a result, there is a crucial need to find new drugs to circumvent resistance problems. In the quest to identify new antimalarial agents a huge number of plant-derived compounds (PDCs) have been investigated. Surprisingly in the in silico PDC screening programs, toxicity filters are either never used or so simple that their interest is limited. In this context, the goal of this study was to show how to take advantage of validated toxicity QSAR models for refining the selection of PDCs. From an original data set of 507 PDCs collected from the literature, the use of toxicity filters for endocrine disruption, developmental toxicity, and hepatotoxicity in conjunction with classical pharmacokinetic filters allowed us to obtain a list of 31 compounds of potential interest. The pros and cons of such a strategy have been discussed.

Designing around Structural Alerts in Drug Discovery

Cumulative research over several decades has implicated the involvement of reactive metabolites in many idiosyncratic adverse drug reactions (IADRs). Consequently, "avoidance" strategies have been inserted into drug discovery paradigms, which include the exclusion of structural alerts and possible termination of reactive metabolite-positive compounds. Several noteworthy examples where reactive metabolite-related liabilities have been resolved through structure-metabolism studies are presented herein. Considerable progress has also been made in addressing the limitations of the avoidance strategy and further refining the process of managing reactive metabolite issues in drug development. These efforts primarily stemmed from the observation that numerous drugs, which contain structural alerts and/or form reactive metabolites, are devoid of ADRs. The Perspective also dwells into an analysis of the structural alert/reactive metabolite concept with a discussion of risk mitigation tactics to support the progression of reactive metabolite-positive drug candidates.

Molecular Informatics Studies of the Iron-Dependent Regulator (ideR) Reveal Potential Novel Anti-Mycobacterium ulcerans Natural Product-Derived Compounds

Buruli ulcer is a neglected tropical disease caused by the bacterium Mycobacterium ulcerans. Its virulence is attributed to the dermo-necrotic polyketide toxin mycolactone, whose synthesis is regressed when its iron acquisition system regulated by the iron-dependent regulator (ideR) is deactivated. Interfering with the activation mechanism of ideR to inhibit the toxin’s synthesis could serve as a possible cure for Buruli ulcer. The three-dimensional structure of the ideR for Mycobacterium ulcerans was generated using homology modeling. A library of 832 African natural products (AfroDB), as well as five known anti-mycobacterial compounds were docked against the metal binding site of the ideR. The area under the curve (AUC) values greater than 0.7 were obtained for the computed Receiver Operating Characteristics (ROC) curves, validating the docking protocol. The identified top hits were pharmacologically profiled using Absorption, Distribution, Metabolism, Elimination and Toxicity (ADMET) predictions and their binding mechanisms were characterized. Four compounds with ZINC IDs ZINC000018185774, ZINC000095485921, ZINC000014417338 and ZINC000005357841 emerged as leads with binding energies of −7.7 kcal/mol, −7.6 kcal/mol, −8.0 kcal/mol and −7.4 kcal/mol, respectively. Induced Fit Docking (IFD) was also performed to account for the protein’s flexibility upon ligand binding and to estimate the best plausible conformation of the complexes. Results obtained from the IFD were consistent with that of the molecular docking with the lead compounds forming interactions with known essential residues and some novel critical residues Thr14, Arg33 and Asp17. A hundred nanoseconds molecular dynamic simulations of the unbound ideR and its complexes with the respective lead compounds revealed changes in the ideR’s conformations induced by ZINC000018185774. Comparison of the lead compounds to reported potent inhibitors by docking them against the DNA-binding domain of the protein also showed the lead compounds to have very close binding affinities to those of the potent inhibitors. Interestingly, structurally similar compounds to ZINC000018185774 and ZINC000014417338, as well as analogues of ZINC000095485921, including quercetin are reported to possess anti-mycobacterial activity. Also, ZINC000005357841 was predicted to possess anti-inflammatory and anti-oxidative activities, which are relevant in Buruli ulcer and iron acquisition mechanisms, respectively. The leads are molecular templates which may serve as essential scaffolds for the design of future anti-mycobacterium ulcerans agents.

Establishing synthesis pathway-host compatibility via enzyme solubility

Current pathway synthesis tools identify possible pathways that can be added to a host to produce the desired target molecule through the exploration of abstract metabolic and reaction network space. However, not many of these tools explore gene-level information required to physically realize the identified synthesis pathways, and none explore enzyme-host compatibility. Developing tools that address this disconnect between abstract reactions/metabolic design space and physical genetic sequence design space will enable expedited experimental efforts that avoid exploring unprofitable synthesis pathways. This work describes a workflow, termed Probabilistic Pathway Assembly with Solubility Confidence Scores (ProPASS), which links synthesis pathway construction with the exploration of the physical design space as imposed by the availability of enzymes with predicted characterized activities within the host. Predicted protein solubility propensity scores are used as a confidence level to quantify the compatibility of each pathway enzyme with the host Escherichia coli (E. coli). This study also presents a database, termed Protein Solubility Database (ProSol DB), which provides solubility confidence scores in E. coli for 240,016 characterized enzymes obtained from UniProtKB/Swiss-Prot. The utility of ProPASS is demonstrated by generating genetic implementations of heterologous synthesis pathways in E. coli that target several commercially useful biomolecules.

The Use of In Silico Models Within a Large Pharmaceutical Company

The present contribution describes how in silico models are applied at different stages of the drug discovery process in the pharmaceutical industry. A thorough description of the most relevant computational methods and tools is given along with an in-depth evaluation of their performance in the context of potential genotoxic impurities assessment.The challenges of predicting the outcome of highly complex studies are discussed followed by considerations on how novel ways to manage, store, share and analyze data may advance knowledge and facilitate modeling efforts.

Literature Review of (Q)SAR Modelling of Nanomaterial Toxicity

Despite the clear benefits that nanotechnology can bring to various sectors of industry, there are serious concerns about the potential health risks associated with engineered nanomaterials (ENMs), intensified by the limited understanding of what makes ENMs toxic and how to make them safe. As the use of ENMs for commercial purposes and the number of workers/end-users being exposed to these materials on a daily basis increases, the need for assessing the potential adverse effects of multifarious ENMs in a time- and cost-effective manner becomes more apparent. One strategy to alleviate the problem of testing a large number and variety of ENMs in terms of their toxicological properties is through the development of computational models that decode the relationships between the physicochemical features of ENMs and their toxicity. Such data-driven models can be used for hazard screening, early identification of potentially harmful ENMs and the toxicity-governing physicochemical properties, and accelerating the decision-making process by maximising the use of existing data. Moreover, these models can also support industrial, regulatory and public needs for designing inherently safer ENMs. This chapter is mainly concerned with the investigation of the applicability of (quantitative) structure-activity relationship ((Q)SAR) methods to modelling of ENMs' toxicity. It summarizes the key components required for successful application of data-driven toxicity prediction techniques to ENMs, the published studies in this field and the current limitations of this approach.

Metabolic pathway reconstruction strategies for central metabolism and natural product biosynthesis

Metabolic pathway reconstruction presents a challenge for understanding metabolic pathways in organisms of interest. Different strategies, i.e., reference-based vs. de novo, must be used for pathway reconstruction depending on the availability of well-characterized enzymatic reactions. If at least one enzyme is already known to catalyze a reaction, its amino acid sequence can be used as a reference for identifying homologous enzymes in the genome of an organism of interest. Where there is no known enzyme able to catalyze a corresponding reaction, however, the reaction and the corresponding enzyme must be predicted de novo from chemical transformations of the putative substrate-product pair. This review summarizes studies involving reference-based and de novo metabolic pathway reconstruction and discusses the importance of the classification and structure-function relationships of enzymes.

Pharmacophore modeling and in silico toxicity assessment of potential anticancer agents from African medicinal plants

Molecular modeling has been employed in the search for lead compounds of chemotherapy to fight cancer. In this study, pharmacophore models have been generated and validated for use in virtual screening protocols for eight known anticancer drug targets, including tyrosine kinase, protein kinase B β, cyclin-dependent kinase, protein farnesyltransferase, human protein kinase, glycogen synthase kinase, and indoleamine 2,3-dioxygenase 1. Pharmacophore models were validated through receiver operating characteristic and Güner–Henry scoring methods, indicating that several of the models generated could be useful for the identification of potential anticancer agents from natural product databases. The validated pharmacophore models were used as three-dimensional search queries for virtual screening of the newly developed AfroCancer database (~400 compounds from African medicinal plants), along with the Naturally Occurring Plant-based Anticancer Compound-Activity-Target dataset (comprising ~1,500 published naturally occurring plant-based compounds from around the world). Additionally, an in silico assessment of toxicity of the two datasets was carried out by the use of 88 toxicity end points predicted by the Lhasa’s expert knowledge-based system (Derek), showing that only an insignificant proportion of the promising anticancer agents would be likely showing high toxicity profiles. A diversity study of the two datasets, carried out using the analysis of principal components from the most important physicochemical properties often used to access drug-likeness of compound datasets, showed that the two datasets do not occupy the same chemical space.

Simultaneous prediction of enzyme orthologs from chemical transformation patterns for de novo metabolic pathway reconstruction

MOTIVATION

Metabolic pathways are an important class of molecular networks consisting of compounds, enzymes and their interactions. The understanding of global metabolic pathways is extremely important for various applications in ecology and pharmacology. However, large parts of metabolic pathways remain unknown, and most organism-specific pathways contain many missing enzymes.

RESULTS

In this study we propose a novel method to predict the enzyme orthologs that catalyze the putative reactions to facilitate the de novo reconstruction of metabolic pathways from metabolome-scale compound sets. The algorithm detects the chemical transformation patterns of substrate-product pairs using chemical graph alignments, and constructs a set of enzyme-specific classifiers to simultaneously predict all the enzyme orthologs that could catalyze the putative reactions of the substrate-product pairs in the joint learning framework. The originality of the method lies in its ability to make predictions for thousands of enzyme orthologs simultaneously, as well as its extraction of enzyme-specific chemical transformation patterns of substrate-product pairs. We demonstrate the usefulness of the proposed method by applying it to some ten thousands of metabolic compounds, and analyze the extracted chemical transformation patterns that provide insights into the characteristics and specificities of enzymes. The proposed method will open the door to both primary (central) and secondary metabolism in genomics research, increasing research productivity to tackle a wide variety of environmental and public health matters.

AVAILABILITY AND IMPLEMENTATION

CONTACT

: maskot@bio.titech.ac.jp.

In silico toxicology: computational methods for the prediction of chemical toxicity

Determining the toxicity of chemicals is necessary to identify their harmful effects on humans, animals, plants, or the environment. It is also one of the main steps in drug design. Animal models have been used for a long time for toxicity testing. However, in vivo animal tests are constrained by time, ethical considerations, and financial burden. Therefore, computational methods for estimating the toxicity of chemicals are considered useful. In silico toxicology is one type of toxicity assessment that uses computational methods to analyze, simulate, visualize, or predict the toxicity of chemicals. In silico toxicology aims to complement existing toxicity tests to predict toxicity, prioritize chemicals, guide toxicity tests, and minimize late-stage failures in drugs design. There are various methods for generating models to predict toxicity endpoints. We provide a comprehensive overview, explain, and compare the strengths and weaknesses of the existing modeling methods and algorithms for toxicity prediction with a particular (but not exclusive) emphasis on computational tools that can implement these methods and refer to expert systems that deploy the prediction models. Finally, we briefly review a number of new research directions in in silico toxicology and provide recommendations for designing in silico models. WIREs Comput Mol Sci 2016, 6:147-172. doi: 10.1002/wcms.1240 For further resources related to this article, please visit the WIREs website.

Systems Pharmacology in Small Molecular Drug Discovery

Drug discovery is a risky, costly and time-consuming process depending on multidisciplinary methods to create safe and effective medicines. Although considerable progress has been made by high-throughput screening methods in drug design, the cost of developing contemporary approved drugs did not match that in the past decade. The major reason is the late-stage clinical failures in Phases II and III because of the complicated interactions between drug-specific, human body and environmental aspects affecting the safety and efficacy of a drug. There is a growing hope that systems-level consideration may provide a new perspective to overcome such current difficulties of drug discovery and development. The systems pharmacology method emerged as a holistic approach and has attracted more and more attention recently. The applications of systems pharmacology not only provide the pharmacodynamic evaluation and target identification of drug molecules, but also give a systems-level of understanding the interaction mechanism between drugs and complex disease. Therefore, the present review is an attempt to introduce how holistic systems pharmacology that integrated in silico ADME/T (i.e., absorption, distribution, metabolism, excretion and toxicity), target fishing and network pharmacology facilitates the discovery of small molecular drugs at the system level.

Developing a Physiologically-Based Pharmacokinetic Model Knowledgebase in Support of Provisional Model Construction

Developing physiologically-based pharmacokinetic (PBPK) models for chemicals can be resource-intensive, as neither chemical-specific parameters nor in vivo pharmacokinetic data are easily available for model construction. Previously developed, well-parameterized, and thoroughly-vetted models can be a great resource for the construction of models pertaining to new chemicals. A PBPK knowledgebase was compiled and developed from existing PBPK-related articles and used to develop new models. From 2,039 PBPK-related articles published between 1977 and 2013, 307 unique chemicals were identified for use as the basis of our knowledgebase. Keywords related to species, gender, developmental stages, and organs were analyzed from the articles within the PBPK knowledgebase. A correlation matrix of the 307 chemicals in the PBPK knowledgebase was calculated based on pharmacokinetic-relevant molecular descriptors. Chemicals in the PBPK knowledgebase were ranked based on their correlation toward ethylbenzene and gefitinib. Next, multiple chemicals were selected to represent exact matches, close analogues, or non-analogues of the target case study chemicals. Parameters, equations, or experimental data relevant to existing models for these chemicals and their analogues were used to construct new models, and model predictions were compared to observed values. This compiled knowledgebase provides a chemical structure-based approach for identifying PBPK models relevant to other chemical entities. Using suitable correlation metrics, we demonstrated that models of chemical analogues in the PBPK knowledgebase can guide the construction of PBPK models for other chemicals.

Metabolome-scale de novo pathway reconstruction using regioisomer-sensitive graph alignments

MOTIVATION

Recent advances in mass spectrometry and related metabolomics technologies have enabled the rapid and comprehensive analysis of numerous metabolites. However, biosynthetic and biodegradation pathways are only known for a small portion of metabolites, with most metabolic pathways remaining uncharacterized.

RESULTS

In this study, we developed a novel method for supervised de novo metabolic pathway reconstruction with an improved graph alignment-based approach in the reaction-filling framework. We proposed a novel chemical graph alignment algorithm, which we called PACHA (Pairwise Chemical Aligner), to detect the regioisomer-sensitive connectivities between the aligned substructures of two compounds. Unlike other existing graph alignment methods, PACHA can efficiently detect only one common subgraph between two compounds. Our results show that the proposed method outperforms previous descriptor-based methods or existing graph alignment-based methods in the enzymatic reaction-likeness prediction for isomer-enriched reactions. It is also useful for reaction annotation that assigns potential reaction characteristics such as EC (Enzyme Commission) numbers and PIERO (Enzymatic Reaction Ontology for Partial Information) terms to substrate-product pairs. Finally, we conducted a comprehensive enzymatic reaction-likeness prediction for all possible uncharacterized compound pairs, suggesting potential metabolic pathways for newly predicted substrate-product pairs.

PROXIMAL: a method for Prediction of Xenobiotic Metabolism

Background

Contamination of the environment with bioactive chemicals has emerged as a potential public health risk. These substances that may cause distress or disease in humans can be found in air, water and food supplies. An open question is whether these chemicals transform into potentially more active or toxic derivatives via xenobiotic metabolizing enzymes expressed in the body. We present a new prediction tool, which we call PROXIMAL (Prediction of Xenobiotic Metabolism) for identifying possible transformation products of xenobiotic chemicals in the liver. Using reaction data from DrugBank and KEGG, PROXIMAL builds look-up tables that catalog the sites and types of structural modifications performed by Phase I and Phase II enzymes. Given a compound of interest, PROXIMAL searches for substructures that match the sites cataloged in the look-up tables, applies the corresponding modifications to generate a panel of possible transformation products, and ranks the products based on the activity and abundance of the enzymes involved.

Results

PROXIMAL generates transformations that are specific for the chemical of interest by analyzing the chemical’s substructures. We evaluate the accuracy of PROXIMAL’s predictions through case studies on two environmental chemicals with suspected endocrine disrupting activity, bisphenol A (BPA) and 4-chlorobiphenyl (PCB3). Comparisons with published reports confirm 5 out of 7 and 17 out of 26 of the predicted derivatives for BPA and PCB3, respectively. We also compare biotransformation predictions generated by PROXIMAL with those generated by METEOR and Metaprint2D-react, two other prediction tools.

Conclusions

PROXIMAL can predict transformations of chemicals that contain substructures recognizable by human liver enzymes. It also has the ability to rank the predicted metabolites based on the activity and abundance of enzymes involved in xenobiotic transformation.

Electronic supplementary material

The online version of this article (doi:10.1186/s12918-015-0241-4) contains supplementary material, which is available to authorized users.

Antihypertensive drugs metabolism: an update to pharmacokinetic profiles and computational approaches

Drug discovery and development is a high-risk enterprise that requires significant investments in capital, time and scientific expertise. The studies of xenobiotic metabolism remain as one of the main topics in the research and development of drugs, cosmetics and nutritional supplements.

Antihypertensive drugs are used for the treatment of high blood pressure, which is one the most frequent symptoms of the patients that undergo cardiovascular diseases such as myocardial infraction and strokes. In current cardiovascular disease pharmacology, four drug clusters - Angiotensin Converting Enzyme Inhibitors, Beta-Blockers, Calcium Channel Blockers and Diuretics - cover the major therapeutic characteristics of the most antihypertensive drugs. The pharmacokinetic and specifically the metabolic profile of the antihypertensive agents are intensively studied because of the broad inter-individual variability on plasma concentrations and the diversity on the efficacy response especially due to the P450 dependent metabolic status they present.

Several computational methods have been developed with the aim to: (i) model and better understand the human drug metabolism; and (ii) enhance the experimental investigation of the metabolism of small xenobiotic molecules. The main predictive tools these methods employ are rule-based approaches, quantitative structure metabolism/activity relationships and docking approaches.

This review paper provides detailed metabolic profiles of the major clusters of antihypertensive agents, including their metabolites and their metabolizing enzymes, and it also provides specific information concerning the computational approaches that have been used to predict the metabolic profile of several antihypertensive drugs.

An evaluation of in-house and off-the-shelf in silico models: implications on guidance for mutagenicity assessment

The evaluation of impurities for genotoxicity using in silico models are commonplace and have become accepted by regulatory agencies. Recently, the ICH M7 Step 4 guidance was published and requires two complementary models for genotoxicity assessments. Over the last ten years, many companies have developed their own internal genotoxicity models built using both public and in-house chemical structures and bacterial mutagenicity data. However, the proprietary nature of internal structures prevents sharing of data and the full OECD compliance of such models. This analysis investigated whether using in-house internal compounds for training models is needed and substantially impacts the results of in silico genotoxicity assessments, or whether using commercial-off-the-shelf (COTS) packages such as Derek Nexus or Leadscope provide adequate performance. We demonstrated that supplementation of COTS packages with a Support Vector Machine (SVM) QSAR model trained on combined in-house and public data does, in fact, improve coverage and accuracy, and reduces the number of compounds needing experimental assessment, i.e., the liability load. This result indicates that there is added value in models trained on both internal and public structures and incorporating such models as part of a consensus approach improves the overall evaluation. Lastly, we optimized an in silico consensus decision-making approach utilizing two COTS models and an internal (SVM) model to minimize false negatives.