Computational predictive programs (expert systems) in toxicology

Artificial intelligence (AI)-it's the end of the tox as we know it (and I feel fine)

The rapid progress of AI impacts diverse scientific disciplines, including toxicology, and has the potential to transform chemical safety evaluation. Toxicology has evolved from an empirical science focused on observing apical outcomes of chemical exposure, to a data-rich field ripe for AI integration. The volume, variety and velocity of toxicological data from legacy studies, literature, high-throughput assays, sensor technologies and omics approaches create opportunities but also complexities that AI can help address. In particular, machine learning is well suited to handle and integrate large, heterogeneous datasets that are both structured and unstructured-a key challenge in modern toxicology. AI methods like deep neural networks, large language models, and natural language processing have successfully predicted toxicity endpoints, analyzed high-throughput data, extracted facts from literature, and generated synthetic data. Beyond automating data capture, analysis, and prediction, AI techniques show promise for accelerating quantitative risk assessment by providing probabilistic outputs to capture uncertainties. AI also enables explanation methods to unravel mechanisms and increase trust in modeled predictions. However, issues like model interpretability, data biases, and transparency currently limit regulatory endorsement of AI. Multidisciplinary collaboration is needed to ensure development of interpretable, robust, and human-centered AI systems. Rather than just automating human tasks at scale, transformative AI can catalyze innovation in how evidence is gathered, data are generated, hypotheses are formed and tested, and tasks are performed to usher new paradigms in chemical safety assessment. Used judiciously, AI has immense potential to advance toxicology into a more predictive, mechanism-based, and evidence-integrated scientific discipline to better safeguard human and environmental wellbeing across diverse populations.

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.

QSAR Methods

This chapter introduces the basis of computational chemistry and discusses how computational methods have been extended from physical to biological properties, and toxicology in particular, modeling. Since about three decades, chemical experimentation is more and more replaced by modeling and virtual experimentation, using a large core of mathematics, chemistry, physics, and algorithms. Animal and wet experiments, aimed at providing a standardized result about a biological property, can be mimicked by modeling methods, globally called in silico methods, all characterized by deducing properties starting from the chemical structures. Two main streams of such models are available: models that consider the whole molecular structure to predict a value, namely QSAR (quantitative structure-activity relationships), and models that check relevant substructures to predict a class, namely SAR. The term in silico discovery is applied to chemical design, to computational toxicology, and to drug discovery. Virtual experiments confirm hypotheses, provide data for regulation, and help in designing new chemicals.

Discrimination of typical cyclic compounds and selection of toxicity evaluation bioassays for coal gasification wastewater (CGW) based on toxicity mechanism of actions (MOAs)

This paper originally investigated toxicity discrimination of typical cyclic compounds and bioassays selection on toxicity evaluation for coal gasification wastewater (CGW) effluent with mechanism-oriented investigation. Initially, representative cyclic toxicants were selected and classified with quantitative structure-toxicity relationship (QSTR). Nitrogen heterocyclic compounds (NHCs) and polycyclic aromatic hydrocarbons (PAHs) were basically discriminated as nonpolar narcotics with significant correlation to hydrophobicity (p < 0.05, R² = 0.8668-0.9635), while phenols were regarded as polar narcotics and reactive compounds due to slight correlation to hydrophobicity (p > 0.05, R² < 0.5). Furthermore, specific mechanism of actions (MOAs) to various organisms revealed that phenols were discriminated as critical source of acute toxicity in CGW, with short-term visible and irreversible damage. However, NHCs and PAHs, which exerted accumulation toxicity rather than acute toxicity, might result in potential mutagenicity and unpredictable risk along the food chain. Afterwards, based on species sensitivity to typical toxicants and application in real CGW effluent, non-applicability of Chlorella vulgaris (C. vulgaris) was validated in toxicity evaluation. While Daphnia magna (D. magna) was suggested as a toxicity bioassay in entire effluent due to the highest sensitivity and applicability. Tetrahymena thermophile (T. pyriformis) might be applicable in effluent with low biodegradability due to similar evaluation results (TU = 8.90) to D. magna (TU = 6.67) in aerobic effluent. Finally, the relationship between toxicity and bioavailability based on typical pollutants and model species illustrated necessity for dualism toxicity-biodegradability investigation on CGW.

QSAR Methods

In this chapter, we introduce the basis of computational chemistry and discuss how computational methods have been extended to some biological properties and toxicology, in particular. Since about 20 years, chemical experimentation is more and more replaced by modeling and virtual experimentation, using a large core of mathematics, chemistry, physics, and algorithms. Then we see how animal experiments, aimed at providing a standardized result about a biological property, can be mimicked by new in silico methods. Our emphasis here is on toxicology and on predicting properties through chemical structures. Two main streams of such models are available: models that consider the whole molecular structure to predict a value, namely QSAR (Quantitative Structure Activity Relationships), and models that find relevant substructures to predict a class, namely SAR. The term in silico discovery is applied to chemical design, to computational toxicology, and to drug discovery. We discuss how the experimental practice in biological science is moving more and more toward modeling and simulation. Such virtual experiments confirm hypotheses, provide data for regulation, and help in designing new chemicals.

Structure- and ligand-based drug design of novel p38-alpha MAPK inhibitors in the fight against the Alzheimer's disease

Alzheimer's disease (AD) is characterized microscopically by the presence of amyloid plaques, which are accumulations of beta-amyloid protein inter-neurons, and neurofibrillary tangles formed predominantly by highly phosphorylated forms of the microtubule-associated protein, tau, which form tangled masses that consume neuronal cell body, possibly leading to neuronal dysfunction and ultimately death. p38α mitogen-activated protein kinase (MAPK) has been implicated in both events associated with AD, tau phosphorylation and inflammation. p38α MAPK pathway is activated by a dual phosphorylation at Thr180 and Tyr182 residues. Drug design of p38α MAPK inhibitors is mainly focused on small molecules that compete for Adenosine triphosphate in the catalytic site. Here, we used different approaches of structure- and ligand-based drug design and medicinal chemistry strategies based on a selected p38α MAPK structure deposited in the Protein Data Bank in complex with inhibitor, as well as others reported in literature. As a result of the virtual screening experiments performed here, as well as molecular dynamics, molecular interaction fields studies, shape and electrostatic similarities, activity and toxicity predictions, and pharmacokinetic and physicochemical properties, we have selected 13 compounds that meet the criteria of low or no toxicity potential, good pharmacotherapeutic profile, predicted activities, and calculated values ​​comparable with those obtained for the reference compounds, while maintaining the main interactions observed for the most potent inhibitors.

Computational toxicological investigation on the mechanism and pathways of xenobiotics metabolized by cytochrome P450: a case of BDE-47

Understanding the transformation mechanism and products of xenobiotics catalyzed by cytochrome P450 enzymes (CYPs) is vital to risk assessment. By density functional theory computation with the B3LYP functional, we simulated the reaction of 2,2',4,4'-tetrabromodiphenyl ether (BDE-47) catalyzed by the active species of CYPs (Compound I). The enzymatic and aqueous environments were simulated by the polarizable continuum model. The results reveal that the addition of Compound I to BDE-47 is the rate-determining step. The addition of Compound I to the ipso and nonsubstituted C atoms forms tetrahedral σ-adducts that further transform into epoxides. Hydroxylation of the epoxides leads to hydroxylated polybrominated diphenyl ethers and 2,4-dibromophenol. The addition to the Br-substituted C2 and C4 atoms has a higher barrier than addition to the nonsubstituted C atoms, forming phenoxide and cyclohexadienone which subsequently undergo debromination/hydroxylation. A novel mechanism was identified in which the approach of Compound I to C2 led to formation of a phenoxide and an expelled Br(-) ion. The predicted products were consistent with the metabolites identified by others. As a first attempt to simulate the enzymatic transformation of a polycyclic compound, this study may enlighten a computational method to predict the biotransformation of xenobiotics catalyzed by CYPs.

A knowledge-based clinical toxicology consultant for diagnosing single exposures

OBJECTIVE

Every year, toxic exposures kill 1200 Americans. To aid in the timely diagnosis and treatment of such exposures, this research investigates the feasibility of a knowledge-based system capable of generating differential diagnoses for human exposures involving unknown toxins.

METHODS

Data mining techniques automatically extract prior probabilities and likelihood ratios from a database managed by the Florida Poison Information Center. Using observed clinical effects, the trained system produces a ranked list of plausible toxic exposures. The resulting system was evaluated using 30,152 single exposure cases. In addition, the effects of two filters for refining diagnosis based on a minimum number of exposure cases and a minimum number of clinical effects were also explored.

RESULTS

The system achieved accuracies (calculated as the percentage of exposures correctly identified in top 10% of trained diagnoses) as high as 79.8% when diagnosing by substance and 78.9% when diagnosing by the major and minor categories of toxins.

CONCLUSIONS

The results of this research are modest, yet promising. At this time, no similar systems are currently in use in the United States and it is hoped that these studies will yield an effective medical decision support system for clinical toxicology.

Paradigm shift in toxicity testing and modeling

The limitations of traditional toxicity testing characterized by high-cost animal models with low-throughput readouts, inconsistent responses, ethical issues, and extrapolability to humans call for alternative strategies for chemical risk assessment. A new strategy using in vitro human cell-based assays has been designed to identify key toxicity pathways and molecular mechanisms leading to the prediction of an in vivo response. The emergence of quantitative high-throughput screening (qHTS) technology has proved to be an efficient way to decompose complex toxicological end points to specific pathways of targeted organs. In addition, qHTS has made a significant impact on computational toxicology in two aspects. First, the ease of mechanism of action identification brought about by in vitro assays has enhanced the simplicity and effectiveness of machine learning, and second, the high-throughput nature and high reproducibility of qHTS have greatly improved the data quality and increased the quantity of training datasets available for predictive model construction. In this review, the benefits of qHTS routinely used in the US Tox21 program will be highlighted. Quantitative structure-activity relationships models built on traditional in vivo data and new qHTS data will be compared and analyzed. In conjunction with the transition from the pilot phase to the production phase of the Tox21 program, more qHTS data will be made available that will enrich the data pool for predictive toxicology. It is perceivable that new in silico toxicity models based on high-quality qHTS data will achieve unprecedented reliability and robustness, thus becoming a valuable tool for risk assessment and drug discovery.

Comparison and possible use of in silico tools for carcinogenicity within REACH legislation

Seven in silico models have been used to assess the prediction accuracy of chemical compound carcinogenicity. More than 1500 compounds with experimental values have been used to evaluate the models. Here we review the application of these models for toxicity prediction and their advantages and disadvantages, discussing the different approaches underlying the models and their main critical points. Some models have fewer false negatives while others are better at avoiding false positives. Since carcinogenicity is typically evaluated using a series of studies, identification of a strategy using one, or preferably a battery of in silico models, could reduce the number of animal studies needed.

Predicting drug side-effect profiles: a chemical fragment-based approach

Background

Drug side-effects, or adverse drug reactions, have become a major public health concern. It is one of the main causes of failure in the process of drug development, and of drug withdrawal once they have reached the market. Therefore, in silico prediction of potential side-effects early in the drug discovery process, before reaching the clinical stages, is of great interest to improve this long and expensive process and to provide new efficient and safe therapies for patients.

Results

In the present work, we propose a new method to predict potential side-effects of drug candidate molecules based on their chemical structures, applicable on large molecular databanks. A unique feature of the proposed method is its ability to extract correlated sets of chemical substructures (or chemical fragments) and side-effects. This is made possible using sparse canonical correlation analysis (SCCA). In the results, we show the usefulness of the proposed method by predicting 1385 side-effects in the SIDER database from the chemical structures of 888 approved drugs. These predictions are performed with simultaneous extraction of correlated ensembles formed by a set of chemical substructures shared by drugs that are likely to have a set of side-effects. We also conduct a comprehensive side-effect prediction for many uncharacterized drug molecules stored in DrugBank, and were able to confirm interesting predictions using independent source of information.

Conclusions

The proposed method is expected to be useful in various stages of the drug development process.

Introduction of jumping fragments in combination with QSARs for the assessment of classification in ecotoxicology

Starting from a random set of structures taken from the European Chemical Bureau (ECB) Web site, an estimation of the classification by acute category in ecotoxicology was carried out. This estimation was based on two approaches. One approach consists in starting with global quantitative structure-activity relationship (QSAR) equations, analyzing the results and defining an interpretation in terms of overall results and mode of action. The other starts with the notion of emerging fragments and more specifically with the introduction of a particular concept: the jumping fragments. This publication studies the scopes and limitations of each approach for the classification of the derivatives. A promising combination of the two methods is proposed for the classification and also for bringing new information about the importance, for the ecotoxicity, of specific chemical fragments considered alone or in association with others.

Expert Systems for Toxicity Prediction

Expert systems offer the facility to predict a toxicity endpoint, as well sometimes as additional relevant information, simply by inputting the chemical structure of a compound. There is now a number of expert systems available, mostly on a commercial basis although a few are free to use or download. This chapter discusses nineteen currently available expert systems, and their performances (if known). Published studies of consensus predictions with these expert systems indicate that these give better results than do individual expert systems.

A test set of compounds with Tetrahymena pyriformis toxicities has been run through the two expert systems known to predict these toxicities; the predictions were quite good, with standard errors of prediction of 0.395 and 0.433 log unit. A further test set of compounds with local lymph node assay skin sensitisation data has been run through seven expert systems, and it was found that consensus predictions were better than were those from any individual expert system.

New public QSAR model for carcinogenicity

BACKGROUND

One of the main goals of the new chemical regulation REACH (Registration, Evaluation and Authorization of Chemicals) is to fulfill the gaps in data concerned with properties of chemicals affecting the human health. (Q)SAR models are accepted as a suitable source of information. The EU funded CAESAR project aimed to develop models for prediction of 5 endpoints for regulatory purposes. Carcinogenicity is one of the endpoints under consideration.

RESULTS

Models for prediction of carcinogenic potency according to specific requirements of Chemical regulation were developed. The dataset of 805 non-congeneric chemicals extracted from Carcinogenic Potency Database (CPDBAS) was used. Counter Propagation Artificial Neural Network (CP ANN) algorithm was implemented. In the article two alternative models for prediction carcinogenicity are described. The first model employed eight MDL descriptors (model A) and the second one twelve Dragon descriptors (model B). CAESAR's models have been assessed according to the OECD principles for the validation of QSAR. For the model validity we used a wide series of statistical checks. Models A and B yielded accuracy of training set (644 compounds) equal to 91% and 89% correspondingly; the accuracy of the test set (161 compounds) was 73% and 69%, while the specificity was 69% and 61%, respectively. Sensitivity in both cases was equal to 75%. The accuracy of the leave 20% out cross validation for the training set of models A and B was equal to 66% and 62% respectively. To verify if the models perform correctly on new compounds the external validation was carried out. The external test set was composed of 738 compounds. We obtained accuracy of external validation equal to 61.4% and 60.0%, sensitivity 64.0% and 61.8% and specificity equal to 58.9% and 58.4% respectively for models A and B.

CONCLUSION

Carcinogenicity is a particularly important endpoint and it is expected that QSAR models will not replace the human experts opinions and conventional methods. However, we believe that combination of several methods will provide useful support to the overall evaluation of carcinogenicity. In present paper models for classification of carcinogenic compounds using MDL and Dragon descriptors were developed. Models could be used to set priorities among chemicals for further testing. The models at the CAESAR site were implemented in java and are publicly accessible.

New public QSAR model for carcinogenicity

BACKGROUND

One of the main goals of the new chemical regulation REACH (Registration, Evaluation and Authorization of Chemicals) is to fulfill the gaps in data concerned with properties of chemicals affecting the human health. (Q)SAR models are accepted as a suitable source of information. The EU funded CAESAR project aimed to develop models for prediction of 5 endpoints for regulatory purposes. Carcinogenicity is one of the endpoints under consideration.

RESULTS

Models for prediction of carcinogenic potency according to specific requirements of Chemical regulation were developed. The dataset of 805 non-congeneric chemicals extracted from Carcinogenic Potency Database (CPDBAS) was used. Counter Propagation Artificial Neural Network (CP ANN) algorithm was implemented. In the article two alternative models for prediction carcinogenicity are described. The first model employed eight MDL descriptors (model A) and the second one twelve Dragon descriptors (model B). CAESAR's models have been assessed according to the OECD principles for the validation of QSAR. For the model validity we used a wide series of statistical checks. Models A and B yielded accuracy of training set (644 compounds) equal to 91% and 89% correspondingly; the accuracy of the test set (161 compounds) was 73% and 69%, while the specificity was 69% and 61%, respectively. Sensitivity in both cases was equal to 75%. The accuracy of the leave 20% out cross validation for the training set of models A and B was equal to 66% and 62% respectively. To verify if the models perform correctly on new compounds the external validation was carried out. The external test set was composed of 738 compounds. We obtained accuracy of external validation equal to 61.4% and 60.0%, sensitivity 64.0% and 61.8% and specificity equal to 58.9% and 58.4% respectively for models A and B.

CONCLUSION

Carcinogenicity is a particularly important endpoint and it is expected that QSAR models will not replace the human experts opinions and conventional methods. However, we believe that combination of several methods will provide useful support to the overall evaluation of carcinogenicity. In present paper models for classification of carcinogenic compounds using MDL and Dragon descriptors were developed. Models could be used to set priorities among chemicals for further testing. The models at the CAESAR site were implemented in java and are publicly accessible.

Counter propagation artificial neural network categorical models for prediction of carcinogenicity for non-congeneric chemicals

One of the main goals of the new chemical regulation REACH (Registration, Evaluation and Authorization of Chemicals) is to fill the gaps on the toxicological properties of chemicals that affect human health. Carcinogenicity is one of the endpoints under consideration. The information obtained from (quantitative) structure-activity relationship ((Q)SAR) models is accepted as an alternative solution to avoid expensive and time-consuming animal tests. The reported results were obtained within the framework of the European project 'Computer Assisted Evaluation of industrial chemical Substances According to Regulations (CAESAR)'. In this article, we demonstrate intermediate results for counter propagation artificial neural network (CP ANN) models for the prediction category of the carcinogenic potency using two-dimensional (2D) descriptors from different software programs. A total of 805 non-congeneric chemicals were extracted from the Carcinogenic Potency Database (CPDBAS). The resulting models had prediction accuracies for internal (training) and external (test) sets as high as 91-93% and 68-70%, respectively. The sensitivity and specificity of the test set were 69-73 and 63-72% correspondingly. High specificity is critical in models for regulatory use that are aimed at ensuring public safety. Thus, the errors that give rise to false negatives are much more relevant. We discuss how we can increase the number of correctly predicted carcinogens using the correlation between the threshold and the values of the sensitivity and specificity.

A combination of 3D-QSAR, docking, local-binding energy (LBE) and GRID study of the species differences in the carcinogenicity of benzene derivatives chemicals

A combination of 3D-QSAR, docking, local-binding energy (LBE) and GRID methods was applied as a tool to study and predict the mechanism of action of 100 carcinogenic benzene derivatives. Two 3D-QSAR models were obtained: (i) model of mouse carcinogenicity on the basis of 100 chemicals (model 1) and (ii) model of the differences in mouse and rat carcinogenicity on the basis of 73 compounds (model 2). 3D-QSAR regression maps indicated the important differences in species carcinogenicity, and the molecular positions associated with them. In order to evaluate the role of P450 metabolic process in carcinogenicity, the following approaches were used. The 3D models of CYP2E1 for mouse and rat were built up. A docking study was applied and the important ligand-protein residues interactions and oxidation positions of the molecules were identified. A new approach for quantitative assessment of metabolism pathways was developed, which enabled us to describe the species differences in CYP2E1 metabolism, and how it can be related to differences in the carcinogenic potential for a subset of compounds. The binding energies of the important substituents (local-binding energy-LBE) were calculated, in order to quantitatively demonstrate the contribution of the substituents in metabolic processes. Furthermore, a computational procedure was used for determining energetically favourable binding sites (GRID examination) of the enzymes. The GRID procedure allowed the identification of some important differences, related to species metabolism in CYP2E1. Comparing GRID, 3D-QSAR maps and LBE results, a similarity was identified, indicating a relationship between P450 metabolic processes and the differences in the carcinogenicity.

An Evaluation of the Predictive Ability of the QSAR Software Packages, DEREK, HAZARDEXPERT and TOPKAT, to Describe Chemically-induced Skin Irritation

According to the REACH chemicals legislation, formally adopted by the EU in 2006, Quantitative Structure–Activity Relationships (QSARs) can be used as alternatives to animal testing, which itself poses specific ethical and economical concerns. A critical assessment of the performance of the QSAR models is therefore the first step toward the reliable use of such computational techniques. This article reports the performance of the skin irritation module of three commercially-available software packages: DEREK, HAZARDEXPERT and TOPKAT. Their performances were tested on the basis of data published in the literature, for 116 chemicals. The results of this study show that only TOPKAT was able to predict the irritative potential for the majority of chemicals, whereas DEREK and HAZARDEXPERT could correctly identify only a few irritant substances.

Toxicity study of allelochemical-like pesticides by a combination of 3D-QSAR, docking, Local Binding Energy (LBE) and GRID approaches

3D-QSAR, Docking, Local Binding Energy (LBE) and GRID methods were integrated as a tool for predicting toxicity and studying mechanisms of action. The method was tested on a set of 73 allelochemical-like pesticides, for which acute toxicity (LD(50)) for the rat was available. 3D-QSAR gave a model with high predictive ability and the regression maps indicated the important toxic chemical substituents. Significant ligand-protein residue interactions and oxidation positions in the binding site were found by docking analysis using CYP1A2 homology modelling. The binding energies of the compounds and the important substituents (Local Binding Energy, LBE) were calculated in order to demonstrate quantitatively the substituent contributions in the metabolism and toxicity. The GRID examination identified the CYP1A2 binding pocket feature. Finally, a 3D-QSAR map was compared to the GRID map, showing good overlaps and confirming the important role of CYP1A2 in allelochemical-like compounds toxicity.

A novel logic-based approach for quantitative toxicology prediction

There is a pressing need for accurate in silico methods to predict the toxicity of molecules that are being introduced into the environment or are being developed into new pharmaceuticals. Predictive toxicology is in the realm of structure activity relationships (SAR), and many approaches have been used to derive such SAR. Previous work has shown that inductive logic programming (ILP) is a powerful approach that circumvents several major difficulties, such as molecular superposition, faced by some other SAR methods. The ILP approach reasons with chemical substructures within a relational framework and yields chemically understandable rules. Here, we report a general new approach, support vector inductive logic programming (SVILP), which extends the essentially qualitative ILP-based SAR to quantitative modeling. First, ILP is used to learn rules, the predictions of which are then used within a novel kernel to derive a support-vector generalization model. For a highly heterogeneous dataset of 576 molecules with known fathead minnow fish toxicity, the cross-validated correlation coefficients (R2CV) from a chemical descriptor method (CHEM) and SVILP are 0.52 and 0.66, respectively. The ILP, CHEM, and SVILP approaches correctly predict 55, 58, and 73%, respectively, of toxic molecules. In a set of 165 unseen molecules, the R2 values from the commercial software TOPKAT and SVILP are 0.26 and 0.57, respectively. In all calculations, SVILP showed significant improvements in comparison with the other methods. The SVILP approach has a major advantage in that it uses ILP automatically and consistently to derive rules, mostly novel, describing fragments that are toxicity alerts. The SVILP is a general machine-learning approach and has the potential of tackling many problems relevant to chemoinformatics including in silico drug design.

Automatic extraction of structural alerts for predicting chromosome aberrations of organic compounds

We use the topological sub-structural molecular design (TOPS-MODE) approach to formulate structural alert rules for chromosome aberration (CA) of organic compounds. First, a classification model was developed to group chemicals as active/inactive respect to CA. A procedure for extracting structural information from orthogonalized TOPS-MODE descriptors was then implemented. The contributions of bonds to CA in all the molecules studied were then generated using the orthogonalized classification model. Using this information we propose 22 structural alert rules which are ready to be implemented in expert systems for the automatic prediction of CA. They include, among others, structural alerts for N-nitroso compounds (ureas, urethanes, guanidines, triazines), nitro compounds (aromatic and heteroaromatic), alkyl esters or phosphoric acids, alkyl methanesulfonates, sulphonic acids and sulphonamides, epoxides, aromatic amines, azaphenanthrene hydrocarbons, etc. The chemico-biological analysis of some of the structural alerts found is also carried out showing the potential of TOPS-MODE as a knowledge generator.

Mutagenic probability estimation of chemical compounds by a novel molecular electrophilicity vector and support vector machine

MOTIVATION

Mutagenicity is among the toxicological end points that pose the highest concern. The accelerated pace of drug discovery has heightened the need for efficient prediction methods. Currently, most available tools fall short of the desired degree of accuracy, and can only provide a binary classification. It is of significance to develop a discriminative and informative model for the mutagenicity prediction.

RESULTS

Here we developed a mutagenic probability prediction model addressing the problem, based on datasets covering a large chemical space. A novel molecular electrophilicity vector (MEV) is first devised to represent the structure profile of chemical compounds. An extended support vector machine (SVM) method is then used to derive the posterior probabilistic estimation of mutagenicity from the MEVs of the training set. The results show that our model gives a better performance than TOPKAT (http://www.accelrys.com) and other previously published methods. In addition, a confidence level related to the prediction can be provided, which may help people make more flexible decisions on chemical ordering or synthesis.

AVAILABILITY

The binary program (ZGTOX_1.1) based on our model and samples of input datasets on Windows PC are available at http://dddc.ac.cn/adme upon request from the authors.

Development of an information-intensive structure-activity relationship model and its application to human respiratory chemical sensitizers

Structure-activity relationship (SAR) models are recognized as powerful tools to predict the toxicologic potential of new or untested chemicals and also provide insight into possible mechanisms of toxicity. Models have been based on physicochemical attributes and structural features of chemicals. We describe herein the development of a new SAR modeling algorithm called cat-SAR that is capable of analyzing and predicting chemical activity from divergent biological response data. The cat-SAR program develops chemical fragment-based SAR models from categorical biological response data (e.g. toxicologically active and inactive compounds). The database selected for model development was a published set of chemicals documented to cause respiratory hypersensitivity in humans. Two models were generated that differed only in that one model included explicate hydrogen containing fragments. The predictive abilities of the models were tested using leave-one-out cross-validation tests. One model had a sensitivity of 0.94 and specificity of 0.87 yielding an overall correct prediction of 91%. The second model had a sensitivity of 0.89, specificity of 0.95 and overall correct prediction of 92%. The demonstrated predictive capabilities of the cat-SAR approach, together with its modeling flexibility and design transparency, suggest the potential for its widespread applicability to toxicity prediction and for deriving mechanistic insight into toxicologic effects.

Genetic Programming for the Induction of Decision Trees to Model Ecotoxicity Data

Automatic induction of decision trees and production rules from data to develop structure-activity models for toxicity prediction has recently received much attention, and the majority of methodologies reported in the literature are based upon recursive partitioning employing greedy searches to choose the best splitting attribute and value at each node. These approaches can be successful; however, the greedy search will necessarily miss regions of the search space. Recent literature has demonstrated the applicability of genetic programming to decision tree induction to overcome this problem. This paper presents a variant of this novel approach, using fewer mutation options and a simpler fitness function, demonstrating its utility in inducing decision trees for ecotoxicity data, via a case study of two data sets giving improved accuracy and generalization ability over a popular decision tree inducer.

Derivation and Validation of Toxicophores for Mutagenicity Prediction

Mutagenicity is one of the numerous adverse properties of a compound that hampers its potential to become a marketable drug. Toxic properties can often be related to chemical structure, more specifically, to particular substructures, which are generally identified as toxicophores. A number of toxicophores have already been identified in the literature. This study aims at increasing the current degree of reliability and accuracy of mutagenicity predictions by identifying novel toxicophores from the application of new criteria for toxicophore rule derivation and validation to a considerably sized mutagenicity dataset. For this purpose, a dataset of 4337 molecular structures with corresponding Ames test data (2401 mutagens and 1936 nonmutagens) was constructed. An initial substructure-search of this dataset showed that most mutagens were detected by applying only eight general toxicophores. From these eight, more specific toxicophores were derived and approved by employing chemical and mechanistic knowledge in combination with statistical criteria. A final set of 29 toxicophores containing new substructures was assembled that could classify the mutagenicity of the investigated dataset with a total classification error of 18%. Furthermore, mutagenicity predictions of an independent validation set of 535 compounds were performed with an error percentage of 15%. Since these error percentages approach the average interlaboratory reproducibility error of Ames tests, which is 15%, it was concluded that these toxicophores can be applied to risk assessment processes and can guide the design of chemical libraries for hit and lead optimization.

Combining Unsupervised and Supervised Artificial Neural Networks to PredictAquatic Toxicity

Most quantitative structure-activity relationship (QSAR) models are linear relationships and significant for only a limited domain of compounds. Here we propose a data-driven approach with a flexible combination of unsupervised and supervised neural networks able to predict the toxicity of a large set of different chemicals while still respecting the QSAR postulates. Since QSAR is applicable only to similar compounds, which have similar biological and physicochemical properties, large numbers of compounds are clustered before building local models, and local models are ensembled to obtain the final result. The approach has been used to develop models to predict the fish toxicity of Pimephales promelas and Tetrahymena pyriformis, a protozoan.

In silico prediction of drug toxicity

It is essential, in order to minimise expensive drug failures due to toxicity being found in late development or even in clinical trials, to determine potential toxicity problems as early as possible. In view of the large libraries of compounds now being handled by combinatorial chemistry and high-throughput screening, identification of putative toxicity is advisable even before synthesis. Thus the use of predictive toxicology is called for. A number of in silico approaches to toxicity prediction are discussed. Quantitative structure-activity relationships (QSARs), relating mostly to specific chemical classes, have long been used for this purpose, and exist for a wide range of toxicity endpoints. However, QSARs also exist for the prediction of toxicity of very diverse libraries, although often such QSARs are of the classification type; that is, they predict simply whether or not a compound is toxic, and do not give an indication of the level of toxicity. Examples are given of all of these. A number of expert systems are available for toxicity prediction, most of them covering a range of toxicity endpoints. Those discussed include TOPKAT, CASE, DEREK, HazardExpert, OncoLogic and COMPACT. Comparative tests of the ability of these systems to predict carcinogenicity show that improvement is still needed. The consensus approach is recommended, whereby the results from several prediction systems are pooled.

A multiple in silico program approach for the prediction of mutagenicity from chemical structure

We have conducted an evaluation of three of the most widely used commercial toxicity prediction programs, Toxicity Prediction by Komputer Assisted Technology (TOPKAT), Deductive Estimation of Risk from Existing Knowledge (DEREK) for Windows (DfW) and CASETOX. The three programs were evaluated for their ability to predict Ames test mutagenicity using 520 proprietary drug candidate (Test set 1) and 94 commercial (Test set 2) compounds. The study demonstrates that these three commercially available programs are useful, with limitations in their ability to predict mutagenicity over a wide range of chemical space, i.e. global predictivity. Individually, each of the programs performed at an acceptable level for overall accuracy, i.e. the ability to predict the correct outcome. However, analysis of the predictions indicates that the overall accuracy figure is heavily weighted by the ability of the programs to correctly predict non-mutagens, whereas none of the programs individually performed well in the prediction of novel mutagenic structures, i.e. Ames positive compounds. The performance of these programs' in predicting Ames positive mutagens appeared to be independent of the chemical utility of the compound, i.e. industrial, agricultural or pharmaceutical. The combination of program predictions provided some improvement in overall accuracy, sensitivity and specificity.

Biokinetic and Toxicodynamic Modelling and its Role in Toxicological Research and Risk Assessment

Toxicological risk assessment for chemicals is still mainly based on highly standardised protocols for animal experimentation and exposure assessment. However, developments in our knowledge of general physiology, in chemicobiological interactions and in (computer-supported) modelling, have resulted in a tremendous change in our understanding of the molecular mechanisms underlying the toxicity of chemicals. This permits the development of biologically based models, in which the biokinetics as well as the toxicodynamics of compounds can be described. In this paper, the possibilities are discussed of developing systems in which the systemic (acute and chronic) toxicities of chemicals can be quantified without the heavy reliance on animal experiments. By integrating data derived from different sources, predictions of toxicity can be made. Key elements in this integrated approach are the evaluation of chemical functionalities representing structural alerts for toxic actions, the construction of biokinetic models on the basis of non-animal data (for example, tissue–blood partition coefficients, in vitro biotransformation parameters), tests or batteries of tests for determining basal cytotoxicity, and more-specific tests for evaluating tissue or organ toxicity. It is concluded that this approach is a useful tool for various steps in toxicological hazard and risk assessment, especially for those forms of toxicity for which validated in vitro and other non-animal tests have already been developed.

Tuning neural and fuzzy-neural networks for toxicity modeling

The need for general reliable models for predicting toxicity has led to the use of artificial intelligence. We applied neural and fuzzy-neural networks with the QSAR approach. We underline how the networks have to be tuned on the data sets generally involved in modeling toxicity. This study was conducted on 562 organic compounds in order to establish models for predictive the acute toxicity in fish.

Modeling toxicity by using supervised kohonen neural networks

Counterprogation neural network is shown to be a powerful and suitable tool for the investigation of toxicity. This study mined a data set of 568 chemicals. Two hundred eighty-two objects were used as the training set and 286 as the test set. The final model developed presents high performances on the data set R(2) = 0.83 (R(2) = 0.97 on the training set, R(2) = 0.59 on the test set). This technique distinguishes itself also for the ability to give to the expert two-dimensional maps suitable for the study of the distribution/clustering of the data and the identification of outliers.

The integration of data on physico-chemical properties, in vitro-derived toxicity data and physiologically based kinetic and dynamic as modelling a tool in hazard and risk assessment. A commentary

Toxicity of a compound for an organism is dependent on the route of exposure, the amount (or concentration), the way in which the compound is taken up, distributes and is eliminated from the organism (ADME, kinetics) and the intrinsic properties (reactivity; mode of action, dynamics) of the compound towards the organism. These three elements: exposure, kinetics and dynamics form the basis of hazard and risk evaluations. Developments in our knowledge of the way in which physico-chemical properties of chemicals (on the one side) and physiological processes in the organism (on the other side) determine a compound's toxicity have greatly increased our understanding of toxicological processes and our ability to interpret experimental results. This has now resulted in the development of model systems in which the above-mentioned processes can be described mathematically. Biokinetic modelling is currently of great interest, but the further development of toxicodynamic modelling is equally important. The combination of both allows the estimation of a compound's critical amount/concentration on the critical site of action, which ideally would be the basis for hazard and risk assessments. In vitro systems have been extremely useful in studying the molecular basis of a chemical's biological activity, including its mechanism(s) of toxic action. Other achievements include the prediction of biological reactivity on the basis of a compound's physico-chemical properties and the construction of quantitative structure-activity relationships (QSARs). However, for the incorporation of in vitro-derived data as well as the results of QSARs, kinetic modelling is indispensable. Thus, biokinetic and toxicodynamic modelling are important (if not crucial) tools in toxicological research and there are increasing opportunities to incorporate the results of this work in hazard and risk assessments. Their implementation will allow a much more scientifically-based and a better structured risk assessment, which will be to a much lesser extent relying on animal experimentation.

Modes of action in ecotoxicology: their role in body burdens, species sensitivity, QSARs, and mixture effects

In contrast to the general research attitude in the basic sciences, environmental sciences are often goal-driven and should provide the scientific basis for risk assessment procedures, cleanup, and precautionary measures and finally provide a decision support for policy and management. Hence, the prominent role of mechanistic studies in ecotoxicology is not only to understand the impact of pollutants on living organisms but also to deduce general principles for the categorization and assessment of effects. The goal of this review is, therefore, not to provide an exhaustive coverage of modes of toxic action and their underlying biochemical mechanisms but rather to discuss critically the application of this knowledge in ecotoxicological risk assessment. Knowing the mechanism or, at least the mode of toxic action is indispensable for developing descriptive and predictive models in ecotoxicology. This review seeks to show the crucial role of target sites, interactions with the target site(s), and mechanisms for an adequate and efficient ecotoxicological risk assessment. Emphasis in the discussion is on target effect concentrations (or target occupancy), species selectivity and species sensitivity, time perspective of effect studies, Quantitative Structure-Activity Relationships (QSAR), and mixture toxicity. A particular focus of this review is on multiple mechanisms. Although the illustrative examples were mainly taken from studies in aquatic ecotoxicology, the proposed conceptual approach is also in principle applicable and even particularly useful for soil and sediment systems. Recommendations for further research and developments include the use of internal effect concentrations and target site concentrations in site-specific risk assessment and as a mixture toxicity parameter as well as general considerations for the derivation of mechanistically meaningful QSAR and other predictive models.

AI and SAR approaches for predicting chemical carcinogenicity: survey and status report

A wide variety of artificial intelligence (AI) and structure-activity relationship (SAR) approaches have been applied to tackling the general problem of predicting rodent chemical carcinogenicity. Given the diversity of chemical structures and mechanisms relative to this endpoint, the shared challenge of these approaches is to accurately delineate classes of active chemicals representing distinct biological and chemical mechanism domains, and within those classes determine the structural features and properties responsible for modulating activity. In the following discussion, we present a survey of AI and SAR approaches that have been applied to the prediction of rodent carcinogenicity, and discuss these in general terms and in the context of the results of two organized prediction exercises (PTE-1 and PTE-2) sponsored by the US National Cancer Institute/National Toxicology Program. Most models participating in these exercises were successful in identifying major structural-alerting classes of active carcinogens, but failed in modeling the more subtle modifiers to activity within those classes. In addition, methods that incorporated mechanism-based reasoning or biological data along with structural information outperformed models limited to structural information exclusively. Finally, a few recent carcinogenicity-modeling efforts are presented illustrating progress in tackling some aspects of the carcinogenicity prediction problem. The first example, a QSAR model for predicting carcinogenic potency of aromatic amines, illustrates that success is possible within well-represented classes of carcinogens. From the second example, a newly developed FDA/OTR MultiCASE model for predicting the carcinogenicity of pharmaceuticals, we conclude that the definitions of biological activity and nature of chemicals in the training set are important determinants of the predictive success and specificity/sensitivity characteristics of a derived model.

Toward the design of chemical libraries for mass screening biased against mutagenic compounds

The ability to develop a chemical into a drug depends on multiple factors. Beyond potency and selectivity, ADME/PK and the toxicological profile of the compound play a significant role in its evaluation as a candidate for development. Those factors are being brought into bear earlier in the discovery process and even into the design of libraries for screening. The purpose of our study is the comparative analysis of simple physical characteristics of compounds that have been reported to be mutagens and nonmutagenic ones. The analysis of differences can lead to the development of knowledge-based biases in the libraries designed for massive screening. For each of four Salmonella strains, TA-98, TA-100, TA-1535, and TA-1537, an analysis of the statistical significance of the deviance of the averages for a number of global properties was carried out. The properties studied included parameters, such as topological indices, and bit strings representing the presence or absence of certain chemical moieties. The results suggest that mutagens display a larger number of hydrogen bond acceptor centers for most strains. Moreover, the use of bit strings points to the importance of certain molecular fragments, such a nitro groups, for the outcome of a mutagenicity study. Development of multivariate models based on global molecular properties or bit strings point to a small advantage of the latter for the prediction of mutagenicity. The benefits of the bit strings are in accord with the use of fragment-based approaches for the prediction of carcinogenicity and mutagenicity in methods described in the literature.

Predicting toxicity: a mechanism of action model of chemical mutagenicity

The increasing importance of theoretical studies for predicting toxicology has aroused the interest of many computational chemists. A new approach has been developed, based on studying at the molecular level two potential mechanisms of action that are related to compound mutagenicity. This approach is the first example that considers both the toxicant and the biological target molecules involved in the interaction. Using some calculated descriptors and a simulation of the interaction chemical, compounds can be classified. More important, the approach helps in understanding and explaining both the correct and the incorrect results, and gives a deeper understanding of the toxic mechanisms. The model has been applied to many compounds and the results are compared with experimental results reported for the corresponding Salmonella tests.

Interpretation of quantitative structure-property and -activity relationships

The potential utility of data reduction methods (e.g. principal component analysis) for the analysis of matrices assembled from the related properties of large sets of compounds is discussed by reference to results obtained from solvent polarity scales, ongoing work on solubilities and sweetness properties, and proposed general treatments of toxicities and gas chromatographic retention indices.