Modeling liver-related adverse effects of drugs using knearest neighbor quantitative structure-activity relationship method

Prediction of Drug-Induced Liver Injury: From Molecular Physicochemical Properties and Scaffold Architectures to Machine Learning Approaches

The process of developing new drugs is widely acknowledged as being time-intensive and requiring substantial financial investment. Despite ongoing efforts to reduce time and expenses in drug development, ensuring medication safety remains an urgent problem. One of the major problems involved in drug development is hepatotoxicity, specifically known as drug-induced liver injury (DILI). The popularity of new drugs often poses a significant barrier during development and frequently leads to their recall after launch. In silico methods have many advantages compared with traditional in vivo and in vitro assays. To establish a more precise and reliable prediction model, it is necessary to utilize an extensive and high-quality database consisting of information on drug molecule properties and structural patterns. In addition, we should also carefully select appropriate molecular descriptors that can be used to accurately depict compound characteristics. The aim of this study was to conduct a comprehensive investigation into the prediction of DILI. First, we conducted a comparative analysis of the physicochemical properties of extensively well-prepared DILI-positive and DILI-negative compounds. Then, we used classic substructure dissection methods to identify structural pattern differences between these two different types of chemical molecules. These findings indicate that it is not feasible to establish property or substructure-based rules for distinguishing between DILI-positive and DILI-negative compounds. Finally, we developed quantitative classification models for predicting DILI using the naïve Bayes classifier (NBC) and recursive partitioning (RP) machine learning techniques. The optimal DILI prediction model was obtained using NBC, which combines 21 physicochemical properties, the VolSurf descriptors and the LCFP_10 fingerprint set. This model achieved a global accuracy (GA) of 0.855 and an area under the curve (AUC) of 0.704 for the training set, while the corresponding values were 0.619 and 0.674 for the test set, respectively. Moreover, indicative substructural fragments favorable or unfavorable for DILI were identified from the best naïve Bayesian classification model. These findings may help prioritize lead compounds in the early stage of drug development pipelines.

Predicting Chemical Immunotoxicity through Data-Driven QSAR Modeling of Aryl Hydrocarbon Receptor Agonism and Related Toxicity Mechanisms

Computational modeling has emerged as a time-saving and cost-effective alternative to traditional animal testing for assessing chemicals for their potential hazards. However, few computational modeling studies for immunotoxicity were reported, with few models available for predicting toxicants due to the lack of training data and the complex mechanisms of immunotoxicity. In this study, we employed a data-driven quantitative structure-activity relationship (QSAR) modeling workflow to extensively enlarge the limited training data by revealing multiple targets involved in immunotoxicity. To this end, a probe data set of 6,341 chemicals was obtained from a high-throughput screening (HTS) assay testing for the activation of the aryl hydrocarbon receptor (AhR) signaling pathway, a key event leading to immunotoxicity. Searching this probe data set against PubChem yielded 3,183 assays with testing results for varying proportions of these 6,341 compounds. 100 assays were selected to develop QSAR models based on their correlations to AhR agonism. Twelve individual QSAR models were built for each assay using combinations of four machine-learning algorithms and three molecular fingerprints. 5-fold cross-validation of the resulting models showed good predictivity (average CCR = 0.73). A total of 20 assays were further selected based on QSAR model performance, and their resulting QSAR models showed good predictivity of potential immunotoxicants from external chemicals. This study provides a computational modeling strategy that can utilize large public toxicity data sets for modeling immunotoxicity and other toxicity endpoints, which have limited training data and complicated toxicity mechanisms.

Staying Ahead of the Game: How SARS-CoV-2 has Accelerated the Application of Machine Learning in Pandemic Management

In recent years, machine learning (ML) techniques have garnered considerable interest for their potential use in accelerating the rate of drug discovery. With the emergence of the severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) pandemic, the utilization of ML has become even more crucial in the search for effective antiviral medications. The pandemic has presented the scientific community with a unique challenge, and the rapid identification of potential treatments has become an urgent priority. Researchers have been able to accelerate the process of identifying drug candidates, repurposing existing drugs, and designing new compounds with desirable properties using machine learning in drug discovery. To train predictive models, ML techniques in drug discovery rely on the analysis of large datasets, including both experimental and clinical data. These models can be used to predict the biological activities, potential side effects, and interactions with specific target proteins of drug candidates. This strategy has proven to be an effective method for identifying potential coronavirus disease 2019 (COVID-19) and other disease treatments. This paper offers a thorough analysis of the various ML techniques implemented to combat COVID-19, including supervised and unsupervised learning, deep learning, and natural language processing. The paper discusses the impact of these techniques on pandemic drug development, including the identification of potential treatments, the understanding of the disease mechanism, and the creation of effective and safe therapeutics. The lessons learned can be applied to future outbreaks and drug discovery initiatives.

In silico modeling-based new alternative methods to predict drug and herb-induced liver injury: A review

New approach methods (NAMs) have been developed to predict a wide range of toxicities through innovative technologies. Liver injury is one of the most extensively studied endpoints due to its severity and frequency, occurring among populations that consume drugs or dietary supplements. In this review, we focus on recent developments of in silico modeling for liver injury prediction using deep learning and in vitro data based on adverse outcome pathways (AOPs). Despite these models being mainly developed using datasets generated from drug-like molecules, they were also applied to the prediction of hepatotoxicity caused by herbal products. As deep learning has achieved great success in many different fields, advanced machine learning algorithms have been actively applied to improve the accuracy of in silico models. Additionally, the development of liver AOPs, combined with big data in toxicology, has been valuable in developing in silico models with enhanced predictive performance and interpretability. Specifically, one approach involves developing structure-based models for predicting molecular initiating events of liver AOPs, while others use in vitro data with structure information as model inputs for making predictions. Even though liver injury remains a difficult endpoint to predict, advancements in machine learning algorithms and the expansion of in vitro databases with relevant biological knowledge have made a huge impact on improving in silico modeling for drug-induced liver injury prediction.

Artificial Intelligence in Drug Toxicity Prediction: Recent Advances, Challenges, and Future Perspectives

Toxicity prediction is a critical step in the drug discovery process that helps identify and prioritize compounds with the greatest potential for safe and effective use in humans, while also reducing the risk of costly late-stage failures. It is estimated that over 30% of drug candidates are discarded owing to toxicity. Recently, artificial intelligence (AI) has been used to improve drug toxicity prediction as it provides more accurate and efficient methods for identifying the potentially toxic effects of new compounds before they are tested in human clinical trials, thus saving time and money. In this review, we present an overview of recent advances in AI-based drug toxicity prediction, including the use of various machine learning algorithms and deep learning architectures, of six major toxicity properties and Tox21 assay end points. Additionally, we provide a list of public data sources and useful toxicity prediction tools for the research community and highlight the challenges that must be addressed to enhance model performance. Finally, we discuss future perspectives for AI-based drug toxicity prediction. This review can aid researchers in understanding toxicity prediction and pave the way for new methods of drug discovery.

Prediction of drug-induced hepatotoxicity based on histopathological whole slide images

Liver is an important metabolic organ in human body and is sensitive to toxic chemicals or drugs. Adverse reactions caused by drug hepatotoxicity will damage the liver and hepatotoxicity is the leading cause of removal of approved drugs from the market. Therefore, it is of great significance to identify liver toxicity as early as possible in the drug development process. In this study, we developed a predictive model for drug hepatotoxicity based on histopathological whole slide images (WSI) which are the by-product of drug experiments and have received little attention. To better represent the WSIs, we constructed a graph representation for each WSI by dividing it into small patches, taking sampled patches as nodes and calculating the correlation coefficients between node features as the edges of the graph structure. Then a WSI-level graph convolutional network (GCN) was built to effectively extract the node information of the graph and predict the toxicity. In addition, we introduced a gated attention global context vector (gaGCV) to combine the global context to make node features to contain more comprehensive information. The results validated on rat liver in vivo data from the Open TG-GATES show that the use of WSI for the prediction of toxicity is feasible and effective.

Quantum machine learning for chemistry and physics

Machine learning (ML) has emerged as a formidable force for identifying hidden but pertinent patterns within a given data set with the objective of subsequent generation of automated predictive behavior. In recent years, it is safe to conclude that ML and its close cousin, deep learning (DL), have ushered in unprecedented developments in all areas of physical sciences, especially chemistry. Not only classical variants of ML, even those trainable on near-term quantum hardwares have been developed with promising outcomes. Such algorithms have revolutionized materials design and performance of photovoltaics, electronic structure calculations of ground and excited states of correlated matter, computation of force-fields and potential energy surfaces informing chemical reaction dynamics, reactivity inspired rational strategies of drug designing and even classification of phases of matter with accurate identification of emergent criticality. In this review we shall explicate a subset of such topics and delineate the contributions made by both classical and quantum computing enhanced machine learning algorithms over the past few years. We shall not only present a brief overview of the well-known techniques but also highlight their learning strategies using statistical physical insight. The objective of the review is not only to foster exposition of the aforesaid techniques but also to empower and promote cross-pollination among future research in all areas of chemistry which can benefit from ML and in turn can potentially accelerate the growth of such algorithms.

Detection of Synergistic Interaction on an Additive Scale Between Two Drugs on Abnormal Elevation of Serum Alanine Aminotransferase Using Machine-Learning Algorithms

Drug-induced liver injury (DILI) is a common adverse drug reaction, with abnormal elevation of serum alanine aminotransferase (ALT). Several clinical studies have investigated whether a combination of two drugs alters the reporting frequency of DILI using traditional statistical methods such as multiple logistic regression (MLR), but this model may over-fit the data. This study aimed to detect a synergistic interaction between two drugs on the risk of abnormal elevation of serum ALT in Japanese adult patients using three machine-learning algorithms: MLR, logistic least absolute shrinkage and selection operator (LASSO) regression, and extreme gradient boosting (XGBoost) algorithms. A total of 58,413 patients were extracted from Nihon University School of Medicine's Clinical Data Warehouse and assigned to case (N = 4,152) and control (N = 54,261) groups. The MLR model over-fitted a training set. In the logistic LASSO regression model, three combinations showed relative excess risk due to interaction (RERI) for abnormal elevation of serum ALT: diclofenac and famotidine (RERI 2.427, 95% bootstrap confidence interval 1.226-11.003), acetaminophen and ambroxol (0.540, 0.087-4.625), and aspirin and cilostazol (0.188, 0.135-3.010). Moreover, diclofenac (adjusted odds ratio 1.319, 95% bootstrap confidence interval 1.189-2.821) and famotidine (1.643, 1.332-2.071) individually affected the risk of abnormal elevation of serum ALT. In the XGBoost model, not only the individual effects of diclofenac (feature importance 0.004) and famotidine (0.016), but also the interaction term (0.004) was included in important predictors. Although further study is needed, the combination of diclofenac and famotidine appears to increase the risk of abnormal elevation of serum ALT in the real world.

The prediction approach of drug-induced liver injury: response to the issues of reproducible science of artificial intelligence in real-world applications

In the previous study, we developed the generalized drug-induced liver injury (DILI) prediction model—ResNet18DNN to predict DILI based on multi-source combined DILI dataset and achieved better performance than that of previously published described DILI prediction models. Recently, we were honored to receive the invitation from the editor to response the Letter to Editor by Liu Zhichao, et al. We were glad that our research has attracted the attention of Liu’s team and they has put forward their opinions on our research. In this response to Letter to the Editor, we will respond to these comments.

Developing an artificial intelligence method for screening hepatotoxic compounds in traditional Chinese medicine and Western medicine combination

BACKGROUNDS

Traditional Chinese medicine and Western medicine combination (TCM-WMC) increased the complexity of compounds ingested.

OBJECTIVE

To develop a method for screening hepatotoxic compounds in TCM-WMC based on chemical structures using artificial intelligence (AI) methods.

METHODS

Drug-induced liver injury (DILI) data was collected from the public databases and published literatures. The total dataset formed by DILI data was randomly divided into training set and test set at a ratio of 3:1 approximately. Machine learning models of SGD (Stochastic Gradient Descent), kNN (k-Nearest Neighbor), SVM (Support Vector Machine), NB (Naive Bayes), DT (Decision Tree), RF (Random Forest), ANN (Artificial Neural Network), AdaBoost, LR (Logistic Regression) and one deep learning model (deep belief network, DBN) were adopted to construct models for screening hepatotoxic compounds.

RESULT

Dataset of 2035 hepatotoxic compounds was collected in this research, in which 1505 compounds were as training set and 530 compounds were as test set. Results showed that RF obtained 0.838 of classification accuracy (CA), 0.827 of F1-score, 0.832 of Precision, 0.838 of Recall, 0.814 of area under the curve (AUC) on the training set and 0.767 of CA, 0.731 of F1, 0.739 of Precision, 0.767 of Recall, 0.739 of AUC on the test set, which was better than other eight machine learning methods. The DBN obtained 82.2% accuracy on the test set, which was higher than any other machine learning models on the test set.

CONCLUSION

The DILI AI models were expected to effectively screen hepatotoxic compounds in TCM-WMC.

In Silico Models for Hepatotoxicity

In this chapter, we review the state of the art of predicting human hepatotoxicity using in silico techniques. There has been significant progress in this area over the past 20 years but there are still some challenges ahead. Principally, these challenges are our partial understanding of a very complex biochemical system and our ability to emulate that in a predictive capacity. Here, we provide an overview of the published modeling approaches in this area to date and discuss their design, strengths and weaknesses. It is interesting to note the diversity in modeling approaches, whether they be statistical algorithms or evidenced-based approaches including structural alerts and pharmacophore models. Irrespective of modeling approach, it appears a common theme of access to appropriate, relevant, and high-quality data is a limitation to all and is likely to continue to be the focus of future research.

Unswerving modeling of hepatotoxicity of cadmium containing quantum dots using amalgamation of quasiSMILES, index of ideality of correlation, and consensus modeling

Liver toxicity of quantum dots varies with size, concentration, and other structural as well as experimental parameters. For modeling hepatotoxicity, the eclectic data associated with cadmium containing quantum dots have been used in the creation of quasiSMILES for their representation. The core diameter is normalized for wider applicability and the index of the ideality of correlation is applied to construct the better quantitative features toxicity relationship models. Total eight splits are created and the best model is obtained through split 1 with better prediction criteria of validation set objects. The values of all statistical criteria used in the quality determination of a QSAR model are within the specified range for all the eight toxicity models developed here. Factors like TGA ligand and 0.6-0.7 nm diameter are favorable for liver toxicity while L-cysteine ligand and 0.5-0.6 nm core diameter are helpful in the reduction of toxicity. Further, the intelligent consensus modeling process forms a total of 40 individual and 20 consensus models and the best individual and consensus models are 'Good' in MAE-based criteria. The consensus modeling enhances the prediction ability as well as the accuracy of the developed models and increases the applicability space of the built models for hepatotoxicity prediction of quantum dots.

Joint Decision-Making Model Based on Consensus Modeling Technology for the Prediction of Drug-Induced Liver Injury

Drug-induced liver injury (DILI) is the major cause of clinical trial failure and postmarketing withdrawals of approved drugs. It is very expensive and time-consuming to evaluate hepatotoxicity using animal or cell-based experiments in the early stage of drug development. In this study, an in silico model based on the joint decision-making strategy was developed for DILI assessment using a relatively large dataset of 2608 compounds. Five consensus models were developed with PaDEL descriptors and PubChem, Substructure, Estate, and Klekota–Roth fingerprints, respectively. Submodels for each consensus model were obtained through joint optimization. The parameters and features of each submodel were optimized jointly based on the hybrid quantum particle swarm optimization (HQPSO) algorithm. The application domain (AD) based on the frequency-weighted and distance (FWD)-based method and Tanimoto similarity index showed the wide AD of the qualified consensus models. A joint decision-making model was integrated by the qualified consensus models, and the overwhelming majority principle was used to improve the performance of consensus models. The application scope narrowing caused by the overwhelming majority principle was successfully solved by joint decision-making. The proposed model successfully predicted 99.2% of the compounds in the test set, with an accuracy of 80.0%, a sensitivity of 83.9, and a specificity of 73.3%. For an external validation set containing 390 compounds collected from DILIrank, 98.2% of the compounds were successfully predicted with an accuracy of 79.9%, a sensitivity of 97.1%, and a specificity of 66.0%. Furthermore, 25 privileged substructures responsible for DILI were identified from Substructure, PubChem, and Klekota–Roth fingerprints. These privileged substructures can be regarded as structural alerts in hepatotoxicity evaluation. Compared with the main published studies, our method exhibits certain advantage in data size, transparency, and standardization of the modeling process and accuracy and credibility of prediction results. It is a promising tool for virtual screening in the early stage of drug development.

Machine learning methods to predict the cultivation age of Panacis Quinquefolii Radix

Background

American ginseng (AG) is a valuable medicine widely consumed as a herbal remedy throughout the world. Huge price difference among AG with different growth years leads to intentional adulteration for higher profits. Thus, developing reliable approaches to authenticate the cultivation ages of AG products is of great use in preventing age falsification.

Methods

A total of 106 batches of AG samples along with their 9 physicochemical features were collected and measured from experiments, which was then split into a training set and two test sets (test set 1 and 2) according to the cultivation regions. Principle component analysis (PCA) was carried out to examine the distribution of the three data sets. Four machine learning (ML) algorithms, namely elastic net, k-nearest neighbors, support vector machine and multi-layer perception (MLP) were employed to construct predictive models using the features as inputs and their growth years as outputs. In addition, a similarity-based applicability domain (AD) was defined for these models to ensure the reliability of the predictive results for AG samples produced in different regions.

Results

A positive correlation was observed between the several features and the growth years. PCA revealed diverse distributions among different cultivation regions. The most accurate model derived from MLP shows good prediction power for the fivefold cross validation and the test set 1 with mean square error (MSE) of 0.017 and 0.016 respectively, but a higher MSE value of 1.260 for the test set 2. After applying the AD, all models showed much lower prediction errors for the test samples within AD (IDs) than those outside the AD (ODs). MLP remains the best predictive model with an MSE value of 0.030 for the IDs.

Conclusion

Cultivation years have a close relationship with bioactive components of AG. The constructed models and AD are also able to predict the cultivation years and discriminate samples that have inaccurate prediction results. The AD-equipped models used in this study provide useful tools for determining the age of AG in the market and are freely available at https://github.com/dreadlesss/Panax_age_predictor.

Supplementary Information

The online version contains supplementary material available at 10.1186/s13020-021-00511-5.

Machine learning - Predicting Ames mutagenicity of small molecules

In modern drug discovery, detection of a compound's potential mutagenicity is crucial. However, the traditional method of mutagenicity detection using the Ames test is costly and time consuming as the compounds need to be synthesised and then tested and the results are not always accurate and reproducible. Therefore, it would be advantageous to develop robust in silico models which can accurately predict the mutagenicity of a compound prior to synthesis to overcome the inadequacies of the Ames test. After curation of a previously defined compound mutagenicity library, over 5000 molecules had their chemical fingerprints and molecular properties calculated. Using 8 classification modelling algorithms, including support vector machine (SVM), random forest (RF) and extreme gradient boosting (XGB), a total of 112 predictive models have been constructed. Their performance has been assessed using 10-fold cross validation and a hold-out test set and some of the top performing models have been assessed using the y-randomisation approach. As a result, we have found SVM and XGB models to have good performance during the 10-fold cross validation (AUROC >0.90, sensitivity >0.85, specificity >0.75, balanced accuracy >0.80, Kappa >0.65) and on the test set (AUROC >0.65, sensitivity >0.65, specificity >0.60, balanced accuracy >0.65, Kappa >0.30). We have also identified molecular properties that are the most influential for mutagenicity prediction when combined with chemical molecular fingerprints. Using the Class A mutagenic compounds from the Ames/QSAR International Challenge Project, we were able to verify our models perform better, predicting more mutagens correctly then the StarDrop Ames mutagenicity prediction and TEST mutagenicity prediction.

Impacts of diphenylamine NSAID halogenation on bioactivation risks

Diphenylamine NSAIDs are highly prescribed therapeutics for chronic pain despite causing symptomatic hepatotoxicity through mitochondrial damage in five percent of patients taking them. Differences in toxicity are attributed to structural modifications to the diphenylamine scaffold rather than its inherent toxicity. We hypothesize that marketed diphenylamine NSAID substituents affect preference and efficiency of bioactivation pathways and clearance. We parsed the FDA DILIrank hepatotoxicity database and modeled marketed drug bioactivation into quinone-species metabolites to identify a family of seven clinically relevant diphenylamine NSAIDs. These drugs fell into two subgroups, i.e., acetic acid and propionic acid diphenylamines, varying in hepatotoxicity risks and modeled bioactivation propensities. We carried out steady-state kinetic studies to assess bioactivation pathways by trapping quinone-species metabolites with dansyl glutathione. Analysis of the glutathione adducts by mass spectrometry characterized structures while dansyl fluorescence provided quantitative yields for their formation. Resulting kinetics identified four possible bioactivation pathways among the drugs, but reaction preference and efficiency depended upon structural modifications to the diphenylamine scaffold. Strikingly, diphenylamine dihalogenation promotes formation of quinone metabolites through four distinct metabolic pathways with high efficiency, whereas those without aromatic halogen atoms were metabolized less efficiently through two or fewer metabolic pathways. Overall metabolism of the drugs was comparable with bioactivation accounting for 4-13% of clearance. Lastly, we calculated daily bioload exposure of quinone-species metabolites based on bioactivation efficiency, bioavailability, and maximal daily dose. The results revealed stratification into the two subgroups; propionic acid diphenylamines had an average four-fold greater daily bioload compared to acetic acid diphenylamines. However, the lack of sufficient study on the hepatotoxicity for all drugs prevents further correlative analyses. These findings provide critical insights on the impact of diphenylamine bioactivation as a precursor to hepatotoxicity and thus, provide a foundation for better risk assessment in drug discovery and development.

Machine-Learning Prediction of Oral Drug-Induced Liver Injury (DILI) via Multiple Features and Endpoints

Drug discovery is a costly process which usually takes more than 10 years and billions of dollars for one successful drug to enter the market. Despite all the safety tests, drugs may still cause adverse reactions and be restricted in use or even withdrawn from the market. Drug-induced liver injury (DILI) is one of the major adverse drug reactions, and computational models may be used to predict and reduce it. To assess the computational prediction performance of DILI, we curated DILI endpoints from three databases and prepared drug features including chemical descriptors, therapeutic classifications, gene expressions, and binding proteins. We trained machine-learning models to predict the various DILI endpoints using different drug features. Using the optimal feature sets, the top-performing models obtained areas under the receiver operating characteristic curve (AUC) around 0.8 for some DILI endpoints. We found that some features, including therapeutic classifications and proteins, have good prediction performance towards DILI. We also discovered that the severity of DILI endpoints as well as the selection of negative samples may significantly affect the prediction results. Overall, our study provided a comprehensive collection, curation, and prediction of DILI endpoints using various drug features, which may help the drug researchers to better understand and prevent DILI during the drug discovery process.

Computational Models Using Multiple Machine Learning Algorithms for Predicting Drug Hepatotoxicity with the DILIrank Dataset

Drug-induced liver injury (DILI) remains one of the challenges in the safety profile of both authorized and candidate drugs, and predicting hepatotoxicity from the chemical structure of a substance remains a task worth pursuing. Such an approach is coherent with the current tendency for replacing non-clinical tests with in vitro or in silico alternatives. In 2016, a group of researchers from the FDA published an improved annotated list of drugs with respect to their DILI risk, constituting “the largest reference drug list ranked by the risk for developing drug-induced liver injury in humans” (DILIrank). This paper is one of the few attempting to predict liver toxicity using the DILIrank dataset. Molecular descriptors were computed with the Dragon 7.0 software, and a variety of feature selection and machine learning algorithms were implemented in the R computing environment. Nested (double) cross-validation was used to externally validate the models selected. A total of 78 models with reasonable performance were selected and stacked through several approaches, including the building of multiple meta-models. The performance of the stacked models was slightly superior to other models published. The models were applied in a virtual screening exercise on over 100,000 compounds from the ZINC database and about 20% of them were predicted to be non-hepatotoxic.

In silico Identification and Mechanism Exploration of Hepatotoxic Ingredients in Traditional Chinese Medicine

BACKGROUNDS AND AIMS

Recently, a growing number of hepatotoxicity cases aroused by Traditional Chinese Medicine (TCM) have been reported, causing increasing concern. To date, the reported predictive models for drug induced liver injury show low prediction accuracy and there are still no related reports for hepatotoxicity evaluation of TCM systematically. Additionally, the mechanism of herb induced liver injury (HILI) still remains unknown. The aim of the study was to identify potential hepatotoxic ingredients in TCM and explore the molecular mechanism of TCM against HILI.

MATERIALS AND METHODS

In this study, we developed consensus models for HILI prediction by integrating the best single classifiers. The consensus model with best performance was applied to identify the potential hepatotoxic ingredients from the Traditional Chinese Medicine Systems Pharmacology database (TCMSP). Systems pharmacology analyses, including multiple network construction and KEGG pathway enrichment, were performed to further explore the hepatotoxicity mechanism of TCM.

RESULTS

16 single classifiers were built by combining four machine learning methods with four different sets of fingerprints. After systematic evaluation, the best four single classifiers were selected, which achieved a Matthews correlation coefficient (MCC) value of 0.702, 0.691, 0.659, and 0.717, respectively. To improve the predictive capacity of single models, consensus prediction method was used to integrate the best four single classifiers. Results showed that the consensus model C-3 (MCC = 0.78) outperformed the four single classifiers and other consensus models. Subsequently, 5,666 potential hepatotoxic compounds were identified by C-3 model. We integrated the top 10 hepatotoxic herbs and discussed the hepatotoxicity mechanism of TCM via systems pharmacology approach. Finally, Chaihu was selected as the case study for exploring the molecular mechanism of hepatotoxicity.

CONCLUSION

Overall, this study provides a high accurate approach to predict HILI and an in silico perspective into understanding the hepatotoxicity mechanism of TCM, which might facilitate the discovery and development of new drugs.

An In Silico Model for Predicting Drug-Induced Hepatotoxicity

As one of the leading causes of drug failure in clinical trials, drug-induced liver injury (DILI) seriously impeded the development of new drugs. Assessing the DILI risk of drug candidates in advance has been considered as an effective strategy to decrease the rate of attrition in drug discovery. Recently, there have been continuous attempts in the prediction of DILI. However, it indeed remains a huge challenge to predict DILI successfully. There is an urgent need to develop a quantitative structure-activity relationship (QSAR) model for predicting DILI with satisfactory performance. In this work, we reported a high-quality QSAR model for predicting the DILI risk of xenobiotics by incorporating the use of eight effective classifiers and molecular descriptors provided by Marvin. In model development, a large-scale and diverse dataset consisting of 1254 compounds for DILI was built through a comprehensive literature retrieval. The optimal model was attained by an ensemble method, averaging the probabilities from eight classifiers, with accuracy (ACC) of 0.783, sensitivity (SE) of 0.818, specificity (SP) of 0.748, and area under the receiver operating characteristic curve (AUC) of 0.859. For further validation, three external test sets and a large negative dataset were utilized. Consequently, both the internal and external validation indicated that our model outperformed prior studies significantly. Data provided by the current study will also be a valuable source for modeling/data mining in the future.

In Silico Approaches to Predict Drug-Transporter Interaction Profiles: Data Mining, Model Generation, and Link to Cholestasis

Transport proteins play a crucial role in drug distribution, disposition, and clearance by mediating cellular drug influx and efflux. Inhibition of these transporters may lead to drug-drug interactions or even drug-induced liver injury, such as cholestasis, which comprises a major challenge in drug development process. Thus, computer-based (in silico) models that can predict the pharmacological and toxicological profiles of these small molecules with respect to liver transporters may help in the early prioritization of compounds and hence may lower the high attrition rates. In this chapter, we provide a protocol for in silico prediction of cholestasis by generating validated predictive models. In addition to the two-dimensional molecular descriptors, we include transporter inhibition predictions as descriptors and evaluate the influence of the same on the performance of the cholestasis models.

Three-Level Hepatotoxicity Prediction System Based on Adverse Hepatic Effects

Hepatotoxicity is a major cause of drug withdrawal from the market. To reduce the drug attrition induced by hepatotoxicity, an accurate and efficient hepatotoxicity prediction system must be constructed. In the present study, we constructed a three-level hepatotoxicity prediction system based on different levels of adverse hepatic effects (AHEs) combined with machine learning, using (1) an end point, hepatotoxicity; (2) four hepatotoxicity severity degrees; and (3) specific AHEs. After collecting and curing 15 873 compound-AHE pairs associated with 2017 compounds and 403 AHEs, we constructed 27 models with three end point levels with the random forest algorithm, and obtained accuracies ranging from 67.0 to 78.2% and the area under receiver operating characteristic curves (AUCs) of 0.715-0.875. The 27 models were fully integrated into a tiered hepatotoxicity prediction system. The existence of hepatotoxicity existence, severity degree, and potential AHEs for a given compound could be inferred simultaneously and systematically. Thus, the tiered hepatotoxicity prediction system allows researchers to have significant confidence in confirming compound hepatotoxicity, analyzing hepatotoxicity from multiple perspectives, obtaining warnings for the potential hepatotoxicity severity, and even rapidly selecting the proper in vitro experiments for hepatotoxicity verification. We also applied three external sets (11 drugs or candidates that failed in clinical trials or were withdrawn from the market, the PharmGKB (offsides) database, and an herbal hepatotoxicity data set) to test and validate the prediction ability of our system. Furthermore, the hepatotoxicity prediction system was adapted into a flow framework based on the Konstanz Information Miner, which was made available for researchers.

Determining the Balance Between Drug Efficacy and Safety by the Network and Biological System Profile of Its Therapeutic Target

One of the most challenging puzzles in drug discovery is the identification and characterization of candidate drug of well-balanced profile between efficacy and safety. So far, extensive efforts have been made to evaluate this balance by estimating the quantitative structure–therapeutic relationship and exploring target profile of adverse drug reaction. Particularly, the therapeutic index (TI) has emerged as a key indicator illustrating this delicate balance, and a clinically successful agent requires a sufficient TI suitable for it corresponding indication. However, the TI information are largely unknown for most drugs, and the mechanism underlying the drugs with narrow TI (NTI drugs) is still elusive. In this study, the collective effects of human protein–protein interaction (PPI) network and biological system profile on the drugs' efficacy–safety balance were systematically evaluated. First, a comprehensive literature review of the FDA approved drugs confirmed their NTI status. Second, a popular feature selection algorithm based on artificial intelligence (AI) was adopted to identify key factors differencing the target mechanism between NTI and non-NTI drugs. Finally, this work revealed that the targets of NTI drugs were highly centralized and connected in human PPI network, and the number of similarity proteins and affiliated signaling pathways of the corresponding targets was much higher than those of non-NTI drugs. These findings together with the newly discovered features or feature groups clarified the key factors indicating drug's narrow TI, and could thus provide a novel direction for determining the delicate drug efficacy-safety balance.

Naïve Bayesian Models for Vero Cell Cytotoxicity

PURPOSE

To advance translational research of potential therapeutic small molecules against infectious microbes, the compounds must display a relative lack of mammalian cell cytotoxicity. Vero cell cytotoxicity (CC₅₀) is a common initial assay for this metric. We explored the development of naïve Bayesian models that can enhance the probability of identifying non-cytotoxic compounds.

METHODS

Vero cell cytotoxicity assays were identified in PubChem, reformatted, and curated to create a training set with 8741 unique small molecules. These data were used to develop Bayesian classifiers, which were assessed with internal cross-validation, external tests with a set of 193 compounds from our laboratory, and independent validation with an additional diverse set of 1609 unique compounds from PubChem.

RESULTS

Evaluation with independent, external test and validation sets indicated that cytotoxicity Bayesian models constructed with the ECFP_6 descriptor were more accurate than those that used FCFP_6 fingerprints. The best cytotoxicity Bayesian model displayed predictive power in external evaluations, according to conventional and chance-corrected statistics, as well as enrichment factors.

CONCLUSIONS

The results from external tests demonstrate that our novel cytotoxicity Bayesian model displays sufficient predictive power to help guide translational research. To assist the chemical tool and drug discovery communities, our curated training set is being distributed as part of the Supplementary Material. Graphical Abstract Naive Bayesian models have been trained with publically available data and offer a useful tool for chemical biology and drug discovery to select for small molecules with a high probability of exhibiting acceptably low Vero cell cytotoxicity.

The development and application of in silico models for drug induced liver injury

Drug-induced liver injury (DILI), caused by drugs, herbal agents or nutritional supplements, is a major issue for patients and the pharmaceutical industry. It has been a leading cause of clinical trials failure and withdrawal of FDA approval. In this research, we focused on in silico estimation of chemical DILI potential on humans based on structurally diverse organic chemicals. We developed a series of binary classification models using five different machine learning methods and eight different feature reduction methods. The model, developed with the support vector machine (SVM) and the MACCS fingerprint, performed best both on the test set and external validation. It achieved a prediction accuracy of 80.39% on the test set and 82.78% on external validation. We made this model available at http://opensource.vslead.com/. The user can freely predict the DILI potential of molecules. Furthermore, we analyzed the difference of distributions of 12 key physical-chemical properties between DILI-positive and DILI-negative compounds and 20 privileged substructures responsible for DILI were identified from the Klekota-Roth fingerprint. Moreover, since traditional Chinese medicine (TCM)-induced liver injury is also one of the major concerns among the toxic effects, we evaluated the DILI potential of TCM ingredients using the MACCS_SVM model developed in this study. We hope the model and privileged substructures could be useful complementary tools for chemical DILI evaluation.

Machine Learning Methods in Computational Toxicology

Various methods of machine learning, supervised and unsupervised, linear and nonlinear, classification and regression, in combination with various types of molecular descriptors, both "handcrafted" and "data-driven," are considered in the context of their use in computational toxicology. The use of multiple linear regression, variants of naïve Bayes classifier, k-nearest neighbors, support vector machine, decision trees, ensemble learning, random forest, several types of neural networks, and deep learning is the focus of attention of this review. The role of fragment descriptors, graph mining, and graph kernels is highlighted. The application of unsupervised methods, such as Kohonen's self-organizing maps and related approaches, which allow for combining predictions with data analysis and visualization, is also considered. The necessity of applying a wide range of machine learning methods in computational toxicology is underlined.

Ecotoxicological modelling of cosmetics for aquatic organisms: A QSTR approach

In this study, externally validated quantitative structure-toxicity relationship (QSTR) models were developed for toxicity of cosmetic ingredients on three different ecotoxicologically relevant organisms, namely Pseudokirchneriella subcapitata, Daphnia magna and Pimephales promelas following the OECD guidelines. The final models were developed by partial least squares (PLS) regression technique, which is more robust than multiple linear regression. The obtained model for P. subcapitata shows that molecular size and complexity have significant impacts on the toxicity of cosmetics. In case of P. promelas and D. magna, we found that the largest contribution to the toxicity was shown by hydrophobicity and van der Waals surface area, respectively. All models were validated using both internal and test compounds employing multiple strategies. For each QSTR model, applicability domain studies were also performed using the "Distance to Model in X-space" method. A comparison was made with the ECOSAR predictions in order to prove the good predictive performances of our developed models. Finally, individual models were applied to predict toxicity for an external set of 596 personal care products having no experimental data for at least one of the endpoints, and the compounds were ranked based on a decreasing order of toxicity using a scaling approach.

Predicting drug-induced liver injury: The importance of data curation

Drug-induced liver injury (DILI) is a major issue for both patients and pharmaceutical industry due to insufficient means of prevention/prediction. In the current work we present a 2-class classification model for DILI, generated with Random Forest and 2D molecular descriptors on a dataset of 966 compounds. In addition, predicted transporter inhibition profiles were also included into the models. The initially compiled dataset of 1773 compounds was reduced via a 2-step approach to 966 compounds, resulting in a significant increase (p-value < 0.05) in model performance. The models have been validated via 10-fold cross-validation and against three external test sets of 921, 341 and 96 compounds, respectively. The final model showed an accuracy of 64% (AUC 68%) for 10-fold cross-validation (average of 50 iterations) and comparable values for two test sets (AUC 59%, 71% and 66%, respectively). In the study we also examined whether the predictions of our in-house transporter inhibition models for BSEP, BCRP, P-glycoprotein, and OATP1B1 and 1B3 contributed in improvement of the DILI mode. Finally, the model was implemented with open-source 2D RDKit descriptors in order to be provided to the community as a Python script.

What if the number of nanotoxicity data is too small for developing predictive Nano-QSAR models? An alternative read-across based approach for filling data gaps

Over the past decade, computational nanotoxicology, in particular Quantitative Structure-Activity Relationship models (Nano-QSAR) that help in assessing the biological effects of nanomaterials, have received much attention. In effect, a solid basis for uncovering the relationships between the structure and property/activity of nanoparticles has been created. Nonetheless, six years after the first pioneering computational studies focusing on the investigation of nanotoxicity were commenced, these computational methods still suffer from many limitations. These are mainly related to the paucity of widely available, systematically varied, libraries of experimental data necessary for the development and validation of such models. This results in the still-low acceptance of these methods as valuable research tools for nanosafety and raises the query as to whether these methods could gain wide acceptance of regulatory bodies as alternatives for traditional in vitro methods. This study aimed to give an answer to the following question: How to remedy the paucity of experimental nanotoxicity data and thereby, overcome key roadblock that hinders the development of approaches for data-driven modeling of nanoparticle properties and toxicities? Here, a simple and transparent read-across algorithm for a pre-screening hazard assessment of nanomaterials that provides reasonably accurate results by making the best use of existing limited set of observations will be introduced.

Prediction models for drug-induced hepatotoxicity by using weighted molecular fingerprints

Background

Drug-induced liver injury (DILI) is a critical issue in drug development because DILI causes failures in clinical trials and the withdrawal of approved drugs from the market. There have been many attempts to predict the risk of DILI based on in vivo and in silico identification of hepatotoxic compounds. In the current study, we propose the in silico prediction model predicting DILI using weighted molecular fingerprints.

Results

In this study, we used 881 bits of molecular fingerprint and used as features describing presence or absence of each substructure of compounds. Then, the Bayesian probability of each substructure was calculated and labeled (positive or negative for DILI), and a weighted fingerprint was determined from the ratio of DILI-positive to DILI-negative probability values. Using weighted fingerprint features, the prediction models were trained and evaluated with the Random Forest (RF) and Support Vector Machine (SVM) algorithms. The constructed models yielded accuracies of 73.8% and 72.6%, AUCs of 0.791 and 0.768 in cross-validation. In independent tests, models achieved accuracies of 60.1% and 61.1% for RF and SVM, respectively. The results validated that weighted features helped increase overall performance of prediction models. The constructed models were further applied to the prediction of natural compounds in herbs to identify DILI potential, and 13,996 unique herbal compounds were predicted as DILI-positive with the SVM model.

Conclusions

The prediction models with weighted features increased the performance compared to non-weighted models. Moreover, we predicted the DILI potential of herbs with the best performed model, and the prediction results suggest that many herbal compounds could have potential to be DILI. We can thus infer that taking natural products without detailed references about the relevant pathways may be dangerous. Considering the frequency of use of compounds in natural herbs and their increased application in drug development, DILI labeling would be very important.

Electronic supplementary material

The online version of this article (doi:10.1186/s12859-017-1638-4) contains supplementary material, which is available to authorized users.

Predicting Drug-Induced Cholestasis with the Help of Hepatic Transporters-An in Silico Modeling Approach

Cholestasis represents one out of three types of drug induced liver injury (DILI), which comprises a major challenge in drug development. In this study we applied a two-class classification scheme based on k-nearest neighbors in order to predict cholestasis, using a set of 93 two-dimensional (2D) physicochemical descriptors and predictions of selected hepatic transporters' inhibition (BSEP, BCRP, P-gp, OATP1B1, and OATP1B3). In order to assess the potential contribution of transporter inhibition, we compared whether the inclusion of the transporters' inhibition predictions contributes to a significant increase in model performance in comparison to the plain use of the 93 2D physicochemical descriptors. Our findings were in agreement with literature findings, indicating a contribution not only from BSEP inhibition but a rather synergistic effect deriving from the whole set of transporters. The final optimal model was validated via both 10-fold cross validation and external validation. It performs quite satisfactorily resulting in 0.686 ± 0.013 for accuracy and 0.722 ± 0.014 for area under the receiver operating characteristic curve (AUC) for 10-fold cross-validation (mean ± standard deviation from 50 iterations).

Adverse drug reactions triggered by the common HLA-B*57:01 variant: a molecular docking study

BACKGROUND

Human leukocyte antigen (HLA) surface proteins are directly involved in idiosyncratic adverse drug reactions. Herein, we present a structure-based analysis of the common HLA-B*57:01 variant known to be responsible for several HLA-linked adverse effects such as the abacavir hypersensitivity syndrome.

METHODS

First, we analyzed three X-ray crystal structures involving the HLA-B*57:01 protein variant, the anti-HIV drug abacavir, and different co-binding peptides present in the antigen-binding cleft. We superimposed the three complexes and showed that abacavir had no significant conformational variation whatever the co-binding peptide. Second, we self-docked abacavir in the HLA-B*57:01 antigen binding cleft with and without peptide using Glide. Third, we docked a small test set of 13 drugs with known ADRs and suspected HLA associations.

RESULTS

In the presence of an endogenous co-binding peptide, we found a significant stabilization (~2 kcal/mol) of the docking scores and identified several modified abacavir-peptide interactions indicating that the peptide does play a role in stabilizing the HLA-abacavir complex. Next, our model was used to dock a test set of 13 drugs at HLA-B*57:01 and measured their predicted binding affinities. Drug-specific interactions were observed at the antigen-binding cleft and we were able to discriminate the compounds with known HLA-B*57:01 liability from inactives.

CONCLUSIONS

Overall, our study highlights the relevance of molecular docking for evaluating and analyzing complex HLA-drug interactions. This is particularly important for virtual drug screening over thousands of HLA variants as other experimental techniques (e.g., in vitro HTS) and computational approaches (e.g., molecular dynamics) are more time consuming and expensive to conduct. As the attention for drugs' HLA liability is on the rise, we believe this work participates in encouraging the use of molecular modeling for reliably studying and predicting HLA-drug interactions. Graphical abstract.

Key Challenges and Opportunities Associated with the Use of In Vitro Models to Detect Human DILI: Integrated Risk Assessment and Mitigation Plans

Drug-induced liver injury (DILI) is a major cause of late-stage clinical drug attrition, market withdrawal, black-box warnings, and acute liver failure. Consequently, it has been an area of focus for toxicologists and clinicians for several decades. In spite of considerable efforts, limited improvements in DILI prediction have been made and efforts to improve existing preclinical models or develop new test systems remain a high priority. While prediction of intrinsic DILI has improved, identifying compounds with a risk for idiosyncratic DILI (iDILI) remains extremely challenging because of the lack of a clear mechanistic understanding and the multifactorial pathogenesis of idiosyncratic drug reactions. Well-defined clinical diagnostic criteria and risk factors are also missing. This paper summarizes key data interpretation challenges, practical considerations, model limitations, and the need for an integrated risk assessment. As demonstrated through selected initiatives to address other types of toxicities, opportunities exist however for improvement, especially through better concerted efforts at harmonization of current, emerging and novel in vitro systems or through the establishment of strategies for implementation of preclinical DILI models across the pharmaceutical industry. Perspectives on the incorporation of newer technologies and the value of precompetitive consortia to identify useful practices are also discussed.

Towards cheminformatics-based estimation of drug therapeutic index: Predicting the protective index of anticonvulsants using a new quantitative structure-index relationship approach

The overall efficacy and safety profile of a new drug is partially evaluated by the therapeutic index in clinical studies and by the protective index (PI) in preclinical studies. In-silico predictive methods may facilitate the assessment of these indicators. Although QSAR and QSTR models can be used for predicting PI, their predictive capability has not been evaluated. To test this capability, we developed QSAR and QSTR models for predicting the activity and toxicity of anticonvulsants at accuracy levels above the literature-reported threshold (LT) of good QSAR models as tested by both the internal 5-fold cross validation and external validation method. These models showed significantly compromised PI predictive capability due to the cumulative errors of the QSAR and QSTR models. Therefore, in this investigation a new quantitative structure-index relationship (QSIR) model was devised and it showed improved PI predictive capability that superseded the LT of good QSAR models. The QSAR, QSTR and QSIR models were developed using support vector regression (SVR) method with the parameters optimized by using the greedy search method. The molecular descriptors relevant to the prediction of anticonvulsant activities, toxicities and PIs were analyzed by a recursive feature elimination method. The selected molecular descriptors are primarily associated with the drug-like, pharmacological and toxicological features and those used in the published anticonvulsant QSAR and QSTR models. This study suggested that QSIR is useful for estimating the therapeutic index of drug candidates.

Computational Models for Human and Animal Hepatotoxicity with a Global Application Scope

Hepatic toxicity is a key concern for novel pharmaceutical drugs since it is difficult to anticipate in preclinical models, and it can originate from pharmacologically unrelated drug effects, such as pathway interference, metabolism, and drug accumulation. Because liver toxicity still ranks among the top reasons for drug attrition, the reliable prediction of adverse hepatic effects is a substantial challenge in drug discovery and development. To this end, more effort needs to be focused on the development of improved predictive in-vitro and in-silico approaches. Current computational models often lack applicability to novel pharmaceutical candidates, typically due to insufficient coverage of the chemical space of interest, which is either imposed by size or diversity of the training data. Hence, there is an urgent need for better computational models to allow for the identification of safe drug candidates and to support experimental design. In this context, a large data set comprising 3712 compounds with liver related toxicity findings in humans and animals was collected from various sources. The complex pathology was clustered into 21 preclinical and human hepatotoxicity endpoints, which were organized into three levels of detail. Support vector machine models were trained for each endpoint, using optimized descriptor sets from chemometrics software. The optimized global human hepatotoxicity model has high sensitivity (68%) and excellent specificity (95%) in an internal validation set of 221 compounds. Models for preclinical endpoints performed similarly. To allow for reliable prediction of "truly external" novel compounds, all predictions are tagged with confidence parameters. These parameters are derived from a statistical analysis of the predictive probability densities. The whole approach was validated for an external validation set of 269 proprietary compounds. The models are fully integrated into our early safety in-silico workflow.

A graph-based approach to construct target-focused libraries for virtual screening

Background

Due to exorbitant costs of high-throughput screening, many drug discovery projects commonly employ inexpensive virtual screening to support experimental efforts. However, the vast majority of compounds in widely used screening libraries, such as the ZINC database, will have a very low probability to exhibit the desired bioactivity for a given protein. Although combinatorial chemistry methods can be used to augment existing compound libraries with novel drug-like compounds, the broad chemical space is often too large to be explored. Consequently, the trend in library design has shifted to produce screening collections specifically tailored to modulate the function of a particular target or a protein family.

Methods

Assuming that organic compounds are composed of sets of rigid fragments connected by flexible linkers, a molecule can be decomposed into its building blocks tracking their atomic connectivity. On this account, we developed eSynth, an exhaustive graph-based search algorithm to computationally synthesize new compounds by reconnecting these building blocks following their connectivity patterns.

Results

We conducted a series of benchmarking calculations against the Directory of Useful Decoys, Enhanced database. First, in a self-benchmarking test, the correctness of the algorithm is validated with the objective to recover a molecule from its building blocks. Encouragingly, eSynth can efficiently rebuild more than 80 % of active molecules from their fragment components. Next, the capability to discover novel scaffolds is assessed in a cross-benchmarking test, where eSynth successfully reconstructed 40 % of the target molecules using fragments extracted from chemically distinct compounds. Despite an enormous chemical space to be explored, eSynth is computationally efficient; half of the molecules are rebuilt in less than a second, whereas 90 % take only about a minute to be generated.

Conclusions

eSynth can successfully reconstruct chemically feasible molecules from molecular fragments. Furthermore, in a procedure mimicking the real application, where one expects to discover novel compounds based on a small set of already developed bioactives, eSynth is capable of generating diverse collections of molecules with the desired activity profiles. Thus, we are very optimistic that our effort will contribute to targeted drug discovery. eSynth is freely available to the academic community at www.brylinski.org/content/molecular-synthesis.Graphical abstract