Mixed learning algorithms and features ensemble in hepatotoxicity prediction

Molecular designing of potential environmentally friendly PFAS based on deep learning and generative models

Perfluoroalkyl and polyfluoroalkyl substances (PFAS) are widely used across a spectrum of industrial and consumer goods. Nonetheless, their persistent nature and tendency to accumulate in biological systems pose substantial environmental and health threats. Consequently, striking a balance between maximizing product efficiency and minimizing environmental and health risks by tailoring the molecular structure of PFAS has become a pivotal challenge in the fields of environmental chemistry and sustainable development. To address this issue, a computational workflow was proposed for designing an environmentally friendly PFAS by incorporating deep learning (DL) and molecular generative models. The hybrid DL architecture MolHGT+ based on heterogeneous graph neural network with transformer-like attention was applied to predict the surface tension, bioaccumulation, and hepatotoxicity of the molecules. Through virtual screening of the PFAS master database using MolHGT+, the findings indicate that incorporating the siloxane group and betaine fragment can effectively decrease both the bioaccumulation and hepatotoxicity of PFAS while preserving low surface tension. In addition, molecular generative models were employed to create a structurally diverse pool of novel PFASs with the aforementioned hit molecules serving as the initial template structures. Overall, our study presents a promising AI-driven method for advancing the development of environmentally friendly PFAS.

The application of chemical similarity measures in an unconventional modeling framework c-RASAR along with dimensionality reduction techniques to a representative hepatotoxicity dataset

With the exponential progress in the field of cheminformatics, the conventional modeling approaches have so far been to employ supervised and unsupervised machine learning (ML) and deep learning models, utilizing the standard molecular descriptors, which represent the structural, physicochemical, and electronic properties of a particular compound. Deviating from the conventional approach, in this investigation, we have employed the classification Read-Across Structure-Activity Relationship (c-RASAR), which involves the amalgamation of the concepts of classification-based quantitative structure-activity relationship (QSAR) and Read-Across to incorporate Read-Across-derived similarity and error-based descriptors into a statistical and machine learning modeling framework. ML models developed from these RASAR descriptors use similarity-based information from the close source neighbors of a particular query compound. We have employed different classification modeling algorithms on the selected QSAR and RASAR descriptors to develop predictive models for efficient prediction of query compounds' hepatotoxicity. The predictivity of each of these models was evaluated on a large number of test set compounds. The best-performing model was also used to screen a true external data set. The concepts of explainable AI (XAI) coupled with Read-Across were used to interpret the contributions of the RASAR descriptors in the best c-RASAR model and to explain the chemical diversity in the dataset. The application of various unsupervised dimensionality reduction techniques like t-SNE and UMAP and the supervised ARKA framework showed the usefulness of the RASAR descriptors over the selected QSAR descriptors in their ability to group similar compounds, enhancing the modelability of the dataset and efficiently identifying activity cliffs. Furthermore, the activity cliffs were also identified from Read-Across by observing the nature of compounds constituting the nearest neighbors for a particular query compound. On comparing our simple linear c-RASAR model with the previously reported models developed using the same dataset derived from the US FDA Orange Book ( https://www.accessdata.fda.gov/scripts/cder/ob/index.cfm ), it was observed that our model is simple, reproducible, transferable, and highly predictive. The performance of the LDA c-RASAR model on the true external set supersedes that of the previously reported work. Therefore, the present simple LDA c-RASAR model can efficiently be used to predict the hepatotoxicity of query chemicals.

Comprehensive hepatotoxicity prediction: ensemble model integrating machine learning and deep learning

Background

Chemicals may lead to acute liver injuries, posing a serious threat to human health. Achieving the precise safety profile of a compound is challenging due to the complex and expensive testing procedures. In silico approaches will aid in identifying the potential risk of drug candidates in the initial stage of drug development and thus mitigating the developmental cost.

Methods

In current studies, QSAR models were developed for hepatotoxicity predictions using the ensemble strategy to integrate machine learning (ML) and deep learning (DL) algorithms using various molecular features. A large dataset of 2588 chemicals and drugs was randomly divided into training (80%) and test (20%) sets, followed by the training of individual base models using diverse machine learning or deep learning based on three different kinds of descriptors and fingerprints. Feature selection approaches were employed to proceed with model optimizations based on the model performance. Hybrid ensemble approaches were further utilized to determine the method with the best performance.

Results

The voting ensemble classifier emerged as the optimal model, achieving an excellent prediction accuracy of 80.26%, AUC of 82.84%, and recall of over 93% followed by bagging and stacking ensemble classifiers method. The model was further verified by an external test set, internal 10-fold cross-validation, and rigorous benchmark training, exhibiting much better reliability than the published models.

Conclusion

The proposed ensemble model offers a dependable assessment with a good performance for the prediction regarding the risk of chemicals and drugs to induce liver damage.

Hybrid non-animal modeling: A mechanistic approach to predict chemical hepatotoxicity

Developing mechanistic non-animal testing methods based on the adverse outcome pathway (AOP) framework must incorporate molecular and cellular key events associated with target toxicity. Using data from an in vitro assay and chemical structures, we aimed to create a hybrid model to predict hepatotoxicants. We first curated a reference dataset of 869 compounds for hepatotoxicity modeling. Then, we profiled them against PubChem for existing in vitro toxicity data. Of the 2560 resulting assays, we selected the mitochondrial membrane potential (MMP) assay, a high-throughput screening (HTS) tool that can test chemical disruptors for mitochondrial function. Machine learning was applied to develop quantitative structure-activity relationship (QSAR) models with 2536 compounds tested in the MMP assay for screening new compounds. The MMP assay results, including QSAR model outputs, yielded hepatotoxicity predictions for reference set compounds with a Correct Classification Ratio (CCR) of 0.59. The predictivity improved by including 37 structural alerts (CCR = 0.8). We validated our model by testing 37 reference set compounds in human HepG2 hepatoma cells, and reliably predicting them for hepatotoxicity (CCR = 0.79). This study introduces a novel AOP modeling strategy that combines public HTS data, computational modeling, and experimental testing to predict chemical hepatotoxicity.

InterDILI: interpretable prediction of drug-induced liver injury through permutation feature importance and attention mechanism

Safety is one of the important factors constraining the distribution of clinical drugs on the market. Drug-induced liver injury (DILI) is the leading cause of safety problems produced by drug side effects. Therefore, the DILI risk of approved drugs and potential drug candidates should be assessed. Currently, in vivo and in vitro methods are used to test DILI risk, but both methods are labor-intensive, time-consuming, and expensive. To overcome these problems, many in silico methods for DILI prediction have been suggested. Previous studies have shown that DILI prediction models can be utilized as prescreening tools, and they achieved a good performance. However, there are still limitations in interpreting the prediction results. Therefore, this study focused on interpreting the model prediction to analyze which features could potentially cause DILI. For this, five publicly available datasets were collected to train and test the model. Then, various machine learning methods were applied using substructure and physicochemical descriptors as inputs and the DILI label as the output. The interpretation of feature importance was analyzed by recognizing the following general-to-specific patterns: (i) identifying general important features of the overall DILI predictions, and (ii) highlighting specific molecular substructures which were highly related to the DILI prediction for each compound. The results indicated that the model not only captured the previously known properties to be related to DILI but also proposed a new DILI potential substructural of physicochemical properties. The models for the DILI prediction achieved an area under the receiver operating characteristic (AUROC) of 0.88-0.97 and an area under the Precision-Recall curve (AUPRC) of 0.81-0.95. From this, we hope the proposed models can help identify the potential DILI risk of drug candidates at an early stage and offer valuable insights for drug development.

Usage of model combination in computational toxicology

New Approach Methodologies (NAMs) have ushered in a new era in the field of toxicology, aiming to replace animal testing. However, despite these advancements, they are not exempt from the inherent complexities associated with the study's endpoint. In this review, we have identified three major groups of complexities: mechanistic, chemical space, and methodological. The mechanistic complexity arises from interconnected biological processes within a network that are challenging to model in a single step. In the second group, chemical space complexity exhibits significant dissimilarity between compounds in the training and test series. The third group encompasses algorithmic and molecular descriptor limitations and typical class imbalance problems. To address these complexities, this work provides a guide to the usage of a combination of predictive Quantitative Structure-Activity Relationship (QSAR) models, known as metamodels. This combination of low-level models (LLMs) enables a more precise approach to the problem by focusing on different sub-mechanisms or sub-processes. For mechanistic complexity, multiple Molecular Initiating Events (MIEs) or levels of information are combined to form a mechanistic-based metamodel. Regarding the complexity arising from chemical space, two types of approaches were reviewed to construct a fragment-based chemical space metamodel: those with and without structure sharing. Metamodels with structure sharing utilize unsupervised strategies to identify data patterns and build low-level models for each cluster, which are then combined. For situations without structure sharing due to pharmaceutical industry intellectual property, the use of prediction sharing, and federated learning approaches have been reviewed. Lastly, to tackle methodological complexity, various algorithms are combined to overcome their limitations, diverse descriptors are employed to enhance problem definition and balanced dataset combinations are used to address class imbalance issues (methodological-based metamodels). Remarkably, metamodels consistently outperformed classical QSAR models across all cases, highlighting the importance of alternatives to classical QSAR models when faced with such complexities.

Review of machine learning and deep learning models for toxicity prediction

The ever-increasing number of chemicals has raised public concerns due to their adverse effects on human health and the environment. To protect public health and the environment, it is critical to assess the toxicity of these chemicals. Traditional in vitro and in vivo toxicity assays are complicated, costly, and time-consuming and may face ethical issues. These constraints raise the need for alternative methods for assessing the toxicity of chemicals. Recently, due to the advancement of machine learning algorithms and the increase in computational power, many toxicity prediction models have been developed using various machine learning and deep learning algorithms such as support vector machine, random forest, k-nearest neighbors, ensemble learning, and deep neural network. This review summarizes the machine learning- and deep learning-based toxicity prediction models developed in recent years. Support vector machine and random forest are the most popular machine learning algorithms, and hepatotoxicity, cardiotoxicity, and carcinogenicity are the frequently modeled toxicity endpoints in predictive toxicology. It is known that datasets impact model performance. The quality of datasets used in the development of toxicity prediction models using machine learning and deep learning is vital to the performance of the developed models. The different toxicity assignments for the same chemicals among different datasets of the same type of toxicity have been observed, indicating benchmarking datasets is needed for developing reliable toxicity prediction models using machine learning and deep learning algorithms. This review provides insights into current machine learning models in predictive toxicology, which are expected to promote the development and application of toxicity prediction models in the future.

Drug-induced liver injury and COVID-19: Use of artificial intelligence and the updated Roussel Uclaf Causality Assessment Method in clinical practice

BACKGROUND

Liver injury is a relevant condition in coronavirus disease 2019 (COVID-19) inpatients. Pathophysiology varies from direct infection by virus, systemic inflammation or drug-induced adverse reaction (DILI). DILI detection and monitoring is clinically relevant, as it may contribute to poor prognosis, prolonged hospitalization and increase indirect healthcare costs. Artificial Intelligence (AI) applied in data mining of electronic medical records combining abnormal liver tests, keyword searching tools, and risk factors analysis is a relevant opportunity for early DILI detection by automated algorithms.

AIM

To describe DILI cases in COVID-19 inpatients detected from data mining in electronic medical records (EMR) using AI and the updated Roussel Uclaf Causality Assessment Method (RUCAM).

METHODS

The study was conducted in March 2021 in a hospital in southern Brazil. The NoHarm^© system uses AI to support decision making in clinical pharmacy. Hospital admissions were 100523 during this period, of which 478 met the inclusion criteria. From these, 290 inpatients were excluded due to alternative causes of liver injury and/or due to not having COVID-19. We manually reviewed the EMR of 188 patients for DILI investigation. Absence of clinical information excluded most eligible patients. The DILI assessment causality was possible via the updated RUCAM in 17 patients.

RESULTS

Mean patient age was 53 years (SD ± 18.37; range 22-83), most were male (70%), and admitted to the non-intensive care unit sector (65%). Liver injury pattern was mainly mixed, mean time to normalization of liver markers was 10 d, and mean length of hospitalization was 20.5 d (SD ± 16; range 7-70). Almost all patients recovered from DILI and one patient died of multiple organ failure. There were 31 suspected drugs with the following RUCAM score: Possible (n = 24), probable (n = 5), and unlikely (n = 2). DILI agents in our study were ivermectin, bicalutamide, linezolid, azithromycin, ceftriaxone, amoxicillin-clavulanate, tocilizumab, piperacillin-tazobactam, and albendazole. Lack of essential clinical information excluded most patients. Although rare, DILI is a relevant clinical condition in COVID-19 patients and may contribute to poor prognostics.

CONCLUSION

The incidence of DILI in COVID-19 inpatients is rare and the absence of relevant clinical information on EMR may underestimate DILI rates. Prospects involve creation and validation of alerts for risk factors in all DILI patients based on RUCAM assessment causality, alterations of liver biomarkers and AI and machine learning.

The System of Self-Consistent Models: QSAR Analysis of Drug-Induced Liver Toxicity

Removing a drug-like substance that can cause drug-induced liver injury from the drug discovery process is a significant task for medicinal chemistry. In silico models can facilitate this process. Semi-correlation is an approach to building in silico models representing the prediction in the active (1)-inactive (0) format. The so-called system of self-consistent models has been suggested as an approach for two tasks: (i) building up a model and (ii) estimating its predictive potential. However, this approach has been tested so far for regression models. Here, the approach is applied to building up and estimating a categorical hepatotoxicity model using the CORAL software. This new process yields good results: sensitivity = 0.77, specificity = 0.75, accuracy = 0.76, and Matthew correlation coefficient = 0.51 (all compounds) and sensitivity = 0.83, specificity = 0.81, accuracy = 0.83 and Matthew correlation coefficient = 0.63 (validation set).

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.

Machine learning in the identification, prediction and exploration of environmental toxicology: Challenges and perspectives

Over the past few decades, data-driven machine learning (ML) has distinguished itself from hypothesis-driven studies and has recently received much attention in environmental toxicology. However, the use of ML in environmental toxicology remains in the early stages, with knowledge gaps, technical bottlenecks in data quality, high-dimensional/heterogeneous/small-sample data analysis and model interpretability, and a lack of an in-depth understanding of environmental toxicology. Given the above problems, we review the recent progress in the literature and highlight state-of-the-art toxicological studies using ML (such as learning and predicting toxicity in complicated biosystems and multiple-factor environmental scenarios of long-term and large-scale pollution). Beyond predicting simple biological endpoints by integrating untargeted omics and adverse outcome pathways, ML development should focus on revealing toxicological mechanisms. The integration of data-driven ML with other methods (e.g., omics analysis and adverse outcome pathway frameworks) endows ML with widely promising application in revealing toxicological mechanisms. High-quality databases and interpretable algorithms are urgently needed for toxicology and environmental science. Addressing the core issues and future challenges for ML in this review may narrow the knowledge gap between environmental toxicity and computational science and facilitate the control of environmental risk in the future.

Mechanism-driven modeling of chemical hepatotoxicity using structural alerts and an in vitro screening assay

Traditional experimental approaches to evaluate hepatotoxicity are expensive and time-consuming. As an advanced framework of risk assessment, adverse outcome pathways (AOPs) describe the sequence of molecular and cellular events underlying chemical toxicities. We aimed to develop an AOP that can be used to predict hepatotoxicity by leveraging computational modeling and in vitro assays. We curated 869 compounds with known hepatotoxicity classifications as a modeling set and extracted assay data from PubChem. The antioxidant response element (ARE) assay, which quantifies transcriptional responses to oxidative stress, showed a high correlation to hepatotoxicity (PPV=0.82). Next, we developed quantitative structure-activity relationship (QSAR) models to predict ARE activation for compounds lacking testing results. Potential toxicity alerts were identified and used to construct a mechanistic hepatotoxicity model. For experimental validation, 16 compounds in the modeling set and 12 new compounds were selected and tested using an in-house ARE-luciferase assay in HepG2-C8 cells. The mechanistic model showed good hepatotoxicity predictivity (accuracy = 0.82) for these compounds. Potential false positive hepatotoxicity predictions by only using ARE results can be corrected by incorporating structural alerts and vice versa. This mechanistic model illustrates a potential toxicity pathway for hepatotoxicity, and this strategy can be expanded to develop predictive models for other complex toxicities.

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.

An Algorithm Framework for Drug-Induced Liver Injury Prediction Based on Genetic Algorithm and Ensemble Learning

In the process of drug discovery, drug-induced liver injury (DILI) is still an active research field and is one of the most common and important issues in toxicity evaluation research. It directly leads to the high wear attrition of the drug. At present, there are a variety of computer algorithms based on molecular representations to predict DILI. It is found that a single molecular representation method is insufficient to complete the task of toxicity prediction, and multiple molecular fingerprint fusion methods have been used as model input. In order to solve the problem of high dimensional and unbalanced DILI prediction data, this paper integrates existing datasets and designs a new algorithm framework, Rotation-Ensemble-GA (R-E-GA). The main idea is to find a feature subset with better predictive performance after rotating the fusion vector of high-dimensional molecular representation in the feature space. Then, an Adaboost-type ensemble learning method is integrated into R-E-GA to improve the prediction accuracy. The experimental results show that the performance of R-E-GA is better than other state-of-art algorithms including ensemble learning-based and graph neural network-based methods. Through five-fold cross-validation, the R-E-GA obtains an ACC of 0.77, an F1 score of 0.769, and an AUC of 0.842.

Machine Learning Models for Predicting Liver Toxicity

Liver toxicity is a major adverse drug reaction that accounts for drug failure in clinical trials and withdrawal from the market. Therefore, predicting potential liver toxicity at an early stage in drug discovery is crucial to reduce costs and the potential for drug failure. However, current in vivo animal toxicity testing is very expensive and time consuming. As an alternative approach, various machine learning models have been developed to predict potential liver toxicity in humans. This chapter reviews current advances in the development and application of machine learning models for prediction of potential liver toxicity in humans and discusses possible improvements to liver toxicity prediction.

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.

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.

Predictive Model for Drug-Induced Liver Injury Using Deep Neural Networks Based on Substructure Space

Drug-induced liver injury (DILI) is a major concern for drug developers, regulators, and clinicians. However, there is no adequate model system to assess drug-associated DILI risk in humans. In the big data era, computational models are expected to play a revolutionary role in this field. This study aimed to develop a deep neural network (DNN)-based model using extended connectivity fingerprints of diameter 4 (ECFP4) to predict DILI risk. Each data set for the predictive model was retrieved and curated from DILIrank, LiverTox, and other literature. The best model was constructed through ten iterations of stratified 10-fold cross-validation, and the applicability domain was defined based on integer ECFP4 bits of the training set which represented substructures. For the robustness test, we employed the concept of the endurance level. The best model showed an accuracy of 0.731, a sensitivity of 0.714, and a specificity of 0.750 on the validation data set in the complete applicability domain. The model was further evaluated with four external data sets and attained an accuracy of 0.867 on 15 drugs with DILI cases reported since 2019. Overall, the results suggested that the ECFP4-based DNN model represents a new tool to identify DILI risk for the evaluation of drug safety.

ResNet18DNN: prediction approach of drug-induced liver injury by deep neural network with ResNet18

Drug-induced liver injury (DILI) has always been the focus of clinicians and drug researchers. How to improve the performance of the DILI prediction model to accurately predict liver injury was an urgent problem for researchers in the field of medical research. In order to solve this scientific problem, this research collected a comprehensive and accurate dataset of DILI with high recognition and high quality based on clinically confirmed DILI compound datasets, including 1446 chemical compounds. Then, the residual neural network with 18-layer by using more 5-layer blocks (ResNet18) with deep neural network (ResNet18DNN) model was proposed to predict DILI, which was an improved model for DILI prediction through vectorization of compound structure image. In predicting DILI, the ResNet18DNN learned greatly and outperformed the existing state-of-the-art DILI predictors. The results of DILI prediction model based on ResNet18DNN showed that the AUC (area under the curve), accuracy, recall, precision, F1-score and specificity of the training set were 0.973, 0.992, 0.995, 0.994, 0.995 and 0.975; those of test set were, respectively, 0.958, 0.976, 0.935, 0.947, 0.926 and 0.913, which were better than the performance of previously published described DILI prediction models. This method adopted ResNet18 embedding method to vectorize molecular structure images and the evaluation indicators of Resnet18DNN were obtained after 10 000 iterations. This prediction approach will greatly improve the performance of the predictive model of DILI and provide an accurate and precise early warning method for DILI in drug development and clinical medication.

Prediction of Drug-Induced Liver Toxicity Using SVM and Optimal Descriptor Sets

Drug-induced liver toxicity is one of the significant safety challenges for the patient's health and the pharmaceutical industry. It causes termination of drug candidates in clinical trials and also the retractions of approved drugs from the market. Thus, it is essential to identify hepatotoxic compounds in the initial stages of drug development process. The purpose of this study is to construct quantitative structure activity relationship models using machine learning algorithms and systematical feature selection methods for molecular descriptor sets. The models were built from a large and diverse set of 1253 drug compounds and were validated internally with 10-fold cross-validation. In this study, we applied a variety of feature selection techniques to extract the optimal subset of descriptors as modeling features to improve the prediction performance. Experimental results suggested that the support vector machine-based classifier had achieved a better classification accuracy with reduced molecular descriptors. The final optimal model provides an accuracy of 0.811, a sensitivity of 0.840, a specificity of 0.783 and Mathew's correlation coefficient of 0.623 with an internal validation set. Furthermore, this model outperformed the prior studies while evaluated in both the internal and external test sets. The utilization of distinct optimal molecular descriptors as modeling features produce an in silico model with a superior performance.

Alignment-Free Antimicrobial Peptide Predictors: Improving Performance by a Thorough Analysis of the Largest Available Data Set

In the last two decades, a large number of machine-learning-based predictors for the activities of antimicrobial peptides (AMPs) have been proposed. These predictors differ from one another in the learning method and in the training and testing data sets used. Unfortunately, the training data sets present several drawbacks, such as a low representativeness regarding the experimentally validated AMP space, and duplicated peptide sequences between negative and positive data sets. These limitations give a low confidence to most of the approaches to be used in prospective studies. To address these weaknesses, we propose novel modeling and assessing data sets from the largest experimentally validated nonredundant peptide data set reported to date. From these novel data sets, alignment-free quantitative sequence-activity models (AF-QSAMs) based on Random Forest are created to identify general AMPs and their antibacterial, antifungal, antiparasitic, and antiviral functional types. An applicability domain analysis is carried out to determine the reliability of the predictions obtained, which, to the best of our knowledge, is performed for the first time for AMP recognition. A benchmarking is undertaken between the models proposed and several models from the literature that are freely available in 13 programs (ClassAMP, iAMP-2L, ADAM, MLAMP, AMPScanner v2.0, AntiFP, AMPfun, PEPred-suite, AxPEP, CAMPR3, iAMPpred, APIN, and Meta-iAVP). The models proposed are those with the best performance in all of the endpoints modeled, while most of the methods from the literature have weak-to-random predictive agreements. The models proposed are also assessed through Y-scrambling and repeated k-fold cross-validation tests, demonstrating that the outcomes obtained by them are not given by chance. Three chemometric analyses also confirmed the relevance of the peptides descriptors used in the modeling. Therefore, it can be concluded that the models built by fixing the drawbacks existing in the literature contribute to identifying antibacterial, antifungal, antiparasitic, and antiviral peptides with high effectivity and reliability. Models are freely available via the AMPDiscover tool at https://biocom-ampdiscover.cicese.mx/.

Mechanism-Driven Read-Across of Chemical Hepatotoxicants Based on Chemical Structures and Biological Data

Hepatotoxicity is a leading cause of attrition in the drug development process. Traditional preclinical and clinical studies to evaluate hepatotoxicity liabilities are expensive and time consuming. With the advent of critical advancements in high-throughput screening, there has been a rapid accumulation of in vitro toxicity data available to inform the risk assessment of new pharmaceuticals and chemicals. To this end, we curated and merged all available in vivo hepatotoxicity data obtained from the literature and public resources, which yielded a comprehensive database of 4089 compounds that includes hepatotoxicity classifications. After dividing the original database of chemicals into modeling and test sets, PubChem assay data were automatically extracted using an in-house data mining tool and clustered based on relationships between structural fragments and cellular responses in in vitro assays. The resultant PubChem assay clusters were further investigated. During the cross-validation procedure, the biological data obtained from several assay clusters exhibited high predictivity of hepatotoxicity and these assays were selected to evaluate the test set compounds. The read-across results indicated that if a new compound contained specific identified chemical fragments (ie, Molecular Initiating Event) and showed active responses in the relevant selected PubChem assays, there was potential for the chemical to be hepatotoxic in vivo. Furthermore, several mechanisms that might contribute to toxicity were derived from the modeling results including alterations in nuclear receptor signaling and inhibition of DNA repair. This modeling strategy can be further applied to the investigation of other complex chemical toxicity phenomena (eg, developmental and reproductive toxicities) as well as drug efficacy.

The Promise of AI for DILI Prediction

Drug-induced liver injury (DILI) is a common reason for the withdrawal of a drug from the market. Early assessment of DILI risk is an essential part of drug development, but it is rendered challenging prior to clinical trials by the complex factors that give rise to liver damage. Artificial intelligence (AI) approaches, particularly those building on machine learning, range from random forests to more recent techniques such as deep learning, and provide tools that can analyze chemical compounds and accurately predict some of their properties based purely on their structure. This article reviews existing AI approaches to predicting DILI and elaborates on the challenges that arise from the as yet limited availability of data. Future directions are discussed focusing on rich data modalities, such as 3D spheroids, and the slow but steady increase in drugs annotated with DILI risk labels.

Artificial intelligence in the early stages of drug discovery

Although the use of computational methods within the pharmaceutical industry is well established, there is an urgent need for new approaches that can improve and optimize the pipeline of drug discovery and development. In spite of the fact that there is no unique solution for this need for innovation, there has recently been a strong interest in the use of Artificial Intelligence for this purpose. As a matter of fact, not only there have been major contributions from the scientific community in this respect, but there has also been a growing partnership between the pharmaceutical industry and Artificial Intelligence companies. Beyond these contributions and efforts there is an underlying question, which we intend to discuss in this review: can the intrinsic difficulties within the drug discovery process be overcome with the implementation of Artificial Intelligence? While this is an open question, in this work we will focus on the advantages that these algorithms provide over the traditional methods in the context of early drug discovery.

Predicting Drug-Induced Liver Injury Using Convolutional Neural Network and Molecular Fingerprint-Embedded Features

As a critical issue in drug development and postmarketing safety surveillance, drug-induced liver injury (DILI) leads to failures in clinical trials as well as retractions of on-market approved drugs. Therefore, it is important to identify DILI compounds in the early-stages through in silico and in vivo studies. It is difficult using conventional safety testing methods, since the predictive power of most of the existing frameworks is insufficiently effective to address this pharmacological issue. In our study, we employ a natural language processing (NLP) inspired computational framework using convolutional neural networks and molecular fingerprint-embedded features. Our development set and independent test set have 1597 and 322 compounds, respectively. These samples were collected from previous studies and matched with established chemical databases for structural validity. Our study comes up with an average accuracy of 0.89, Matthews's correlation coefficient (MCC) of 0.80, and an AUC of 0.96. Our results show a significant improvement in the AUC values compared to the recent best model with a boost of 6.67%, from 0.90 to 0.96. Also, based on our findings, molecular fingerprint-embedded featurizer is an effective molecular representation for future biological and biochemical studies besides the application of classic molecular fingerprints.

Ensemble Models Based on QuBiLS-MAS Features and Shallow Learning for the Prediction of Drug-Induced Liver Toxicity: Improving Deep Learning and Traditional Approaches

Drug-induced liver injury (DILI) is a key safety issue in the drug discovery pipeline and a regulatory concern. Thus, many in silico tools have been proposed to improve the hepatotoxicity prediction of organic-type chemicals. Here, classifiers for the prediction of DILI were developed by using QuBiLS-MAS 0-2.5D molecular descriptors and shallow machine learning techniques, on a training set composed of 1075 molecules. The best ensemble model build, E13, was obtained with good statistical parameters for the learning series, namely, the following: accuracy = 0.840, sensibility = 0.890, specificity = 0.761, Matthew's correlation coefficient = 0.660, and area under the ROC curve = 0.904. The model was also satisfactorily evaluated with Y-scrambling test, and repeated k-fold cross-validation and repeated k-holdout validation. In addition, an exhaustive external validation was also carried out by using two test sets and five external test sets, with an average accuracy value equal to 0.854 (±0.062) and a coverage equal to 98.4% according to its applicability domain. A statistical comparison of the performance of the E13 model, with regard to results and tools (e.g., Padel DDPredictor Software, Deep Learning DILIserver, and Vslead) reported in the literature, was also performed. In general, E13 presented the best global performance in all experiments. The sum of the ranking differences procedure provided a very similar grouping pattern to that of the M-ANOVA statistical analysis, where E13 was identified as the best model for DILI predictions. A noncommercial and fully cross-platform software for the DILI prediction was also developed, which is freely available at http://tomocomd.com/apps/ptoxra. This software was used for the screening of seven data sets, containing natural products, leads, toxic materials, and FDA approved drugs, to assess the usefulness of the QSAR models in the DILI labeling of organic substances; it was found that 50-92% of the evaluated molecules are positive-DILI compounds. All in all, it can be stated that the E13 model is a relevant method for the prediction of DILI risk in humans, as it shows the best results among all of the methods analyzed.

Mechanism-based integrated assay systems for the prediction of drug-induced liver injury

Drug-induced liver injury (DILI) can cause hepatic failure and result in drug withdrawal from the market. It has host-related and compound-dependent mechanisms. Preclinical prediction of DILI risk is very challenging and safety assessments based on animals inadequately forecast human DILI risk. In contrast, human-derived in vitro cell culture-based models could improve DILI risk prediction accuracy. Here, we developed and validated an innovative method to assess DILI risk associated with various compounds. Fifty-four marketed and withdrawn drugs classified as DILI risks of "most concern", "less concern", and "no concern" were tested using a combination of four assays addressing mitochondrial injury, intrahepatic lipid accumulation, inhibition of bile canalicular network formation, and bile acid accumulation. Using the inhibitory potencies of the drugs evaluated in these in vitro tests, an algorithm with the highest available DILI risk prediction power was built by artificial neural network (ANN) analysis. It had an overall forecasting accuracy of 73%. We excluded the intrahepatic lipid accumulation assay to avoid overfitting. The accuracy of the algorithm in terms of predicting DILI risks was 62% when it was constructed by ANN but only 49% when it was built by the point-added scoring method. The final algorithm based on three assays made no DILI risk prediction errors such as "most concern " instead of "no concern" and vice-versa. Our mechanistic approach may accurately predict DILI risks associated with numerous candidate drugs.

Hepatotoxicity Modeling Using Counter-Propagation Artificial Neural Networks: Handling an Imbalanced Classification Problem

Drug-induced liver injury is a major concern in the drug development process. Expensive and time-consuming in vitro and in vivo studies do not reflect the complexity of the phenomenon. Complementary to wet lab methods are in silico approaches, which present a cost-efficient method for toxicity prediction. The aim of our study was to explore the capabilities of counter-propagation artificial neural networks (CPANNs) for the classification of an imbalanced dataset related to idiosyncratic drug-induced liver injury and to develop a model for prediction of the hepatotoxic potential of drugs. Genetic algorithm optimization of CPANN models was used to build models for the classification of drugs into hepatotoxic and non-hepatotoxic class using molecular descriptors. For the classification of an imbalanced dataset, we modified the classical CPANN training algorithm by integrating random subsampling into the training procedure of CPANN to improve the classification ability of CPANN. According to the number of models accepted by internal validation and according to the prediction statistics on the external set, we concluded that using an imbalanced set with balanced subsampling in each learning epoch is a better approach compared to using a fixed balanced set in the case of the counter-propagation artificial neural network learning methodology.

Predicting reactive sites with quantum chemical topology: carbonyl additions in multicomponent reactions

The reactivity of an atom within a molecule depends mostly on the way the electron density polarizes reflected in the quadrupole moment of the reactive atom.

Toxicity Prediction Method Based on Multi-Channel Convolutional Neural Network

Molecular toxicity prediction is one of the key studies in drug design. In this paper, a deep learning network based on a two-dimension grid of molecules is proposed to predict toxicity. At first, the van der Waals force and hydrogen bond were calculated according to different descriptors of molecules, and multi-channel grids were generated, which could discover more detail and helpful molecular information for toxicity prediction. The generated grids were fed into a convolutional neural network to obtain the result. A Tox21 dataset was used for the evaluation. This dataset contains more than 12,000 molecules. It can be seen from the experiment that the proposed method performs better compared to other traditional deep learning and machine learning methods.

Predicting Drug-Induced Liver Injury Using Ensemble Learning Methods and Molecular Fingerprints

Drug-induced liver injury (DILI) is a major safety concern in the drug-development process, and various methods have been proposed to predict the hepatotoxicity of compounds during the early stages of drug trials. In this study, we developed an ensemble model using 3 machine learning algorithms and 12 molecular fingerprints from a dataset containing 1241 diverse compounds. The ensemble model achieved an average accuracy of 71.1 ± 2.6%, sensitivity (SE) of 79.9 ± 3.6%, specificity (SP) of 60.3 ± 4.8%, and area under the receiver-operating characteristic curve (AUC) of 0.764 ± 0.026 in 5-fold cross-validation and an accuracy of 84.3%, SE of 86.9%, SP of 75.4%, and AUC of 0.904 in an external validation dataset of 286 compounds collected from the Liver Toxicity Knowledge Base. Compared with previous methods, the ensemble model achieved relatively high accuracy and SE. We also identified several substructures related to DILI. In addition, we provide a web server offering access to our models (http://ccsipb.lnu.edu.cn/toxicity/HepatoPred-EL/).

In Silico Prediction of Drug-Induced Liver Injury Based on Ensemble Classifier Method

Drug-induced liver injury (DILI) is a major factor in the development of drugs and the safety of drugs. If the DILI cannot be effectively predicted during the development of the drug, it will cause the drug to be withdrawn from markets. Therefore, DILI is crucial at the early stages of drug research. This work presents a 2-class ensemble classifier model for predicting DILI, with 2D molecular descriptors and fingerprints on a dataset of 450 compounds. The purpose of our study is to investigate which are the key molecular fingerprints that may cause DILI risk, and then to obtain a reliable ensemble model to predict DILI risk with these key factors. Experimental results suggested that 8 molecular fingerprints are very critical for predicting DILI, and also obtained the best ratio of molecular fingerprints to molecular descriptors. The result of the 5-fold cross-validation of the ensemble vote classifier method obtain an accuracy of 77.25%, and the accuracy of the test set was 81.67%. This model could be used for drug-induced liver injury prediction.

Enhancing Acute Oral Toxicity Predictions by using Consensus Modeling and Algebraic Form-Based 0D-to-2D Molecular Encodes

Quantitative structure-activity relationships (QSAR) are introduced to predict acute oral toxicity (AOT), by using the QuBiLS-MAS (acronym for quadratic, bilinear and N-Linear maps based on graph-theoretic electronic-density matrices and atomic weightings) framework for the molecular encoding. Three training sets were employed to build the models: EPA training set (5931 compounds), EPA-full training set (7413 compounds), and Zhu training set (10 152 compounds). Additionally, the EPA test set (1482 compounds) was used for the validation of the QSAR models built on the EPA training set, while the ProTox (425 compounds) and T3DB (284 compounds) external sets were employed for the assessment of all the models. The k-nearest neighbor, multilayer perceptron, random forest, and support vector machine procedures were employed to build several base (individual) models. The base models with R_EPA-training ≥ 0.75 ( R = correlation coefficient) and MAE_EPA-training ≤ 0.5 (MAE = mean absolute error) were retained to build consensus models. As a result, two consensus models based on the minimum operator and denoted as M19 and M22, as well as a consensus model based on the weighted average operator and denoted as M24, were selected as the best ones for each training set considered. According to the applicability domain (AD) analysis performed, model M19 (built on the EPA training set) has MAE_test-AD = 0.4044, MAE_ProTox-AD = 0.4067 and MAE_T3DB-AD = 0.2586 on the EPA test set, ProTox external set, and T3DB external set, respectively; whereas model M22 (built on the EPA-full set) and model M24 (built on the Zhu set) present MAE_ProTox-AD = 0.3992 and MAE_T3DB-AD = 0.2286, and MAE_ProTox-AD = 0.3773 and MAE_T3DB-AD = 0.2471 on the two external sets accounted for, respectively. These outcomes were compared and statistically validated with respect to 14 QSAR methods (e.g., admetSAR, ProTox-II) from the literature. As a result, model M22 presents the best overall performance. In addition, a retrospective study on 261 withdrawn drugs due to their toxic/side effects was performed, to assess the usefulness of prospectively using the QSAR models proposed in the labeling of chemicals. A comparison with regard to the methods from the literature was also made. As a result, model M22 has the best ability of labeling a compound as toxic according to the globally harmonized system of classification and labeling of chemicals. Therefore, it can be concluded that the models proposed, especially model M22, constitute prominent tools for studying AOT, at providing the best results among all the methods examined. A freely available software was also developed to be used in virtual screening tasks ( http://tomocomd.com/apps/ptoxra ).

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.

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.

Prediction Is a Balancing Act: Importance of Sampling Methods to Balance Sensitivity and Specificity of Predictive Models Based on Imbalanced Chemical Data Sets

Increase in the number of new chemicals synthesized in past decades has resulted in constant growth in the development and application of computational models for prediction of activity as well as safety profiles of the chemicals. Most of the time, such computational models and its application must deal with imbalanced chemical data. It is indeed a challenge to construct a classifier using imbalanced data set. In this study, we analyzed and validated the importance of different sampling methods over non-sampling method, to achieve a well-balanced sensitivity and specificity of a machine learning model trained on imbalanced chemical data. Additionally, this study has achieved an accuracy of 93.00%, an AUC of 0.94, F1 measure of 0.90, sensitivity of 96.00% and specificity of 91.00% using SMOTE sampling and Random Forest classifier for the prediction of Drug Induced Liver Injury (DILI). Our results suggest that, irrespective of data set used, sampling methods can have major influence on reducing the gap between sensitivity and specificity of a model. This study demonstrates the efficacy of different sampling methods for class imbalanced problem using binary chemical data sets.

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.

Advancing Predictive Hepatotoxicity at the Intersection of Experimental, in Silico, and Artificial Intelligence Technologies

Adverse drug reactions, particularly those that result in drug-induced liver injury (DILI), are a major cause of drug failure in clinical trials and drug withdrawals. Hepatotoxicity-mediated drug attrition occurs despite substantial investments of time and money in developing cellular assays, animal models, and computational models to predict its occurrence in humans. Underperformance in predicting hepatotoxicity associated with drugs and drug candidates has been attributed to existing gaps in our understanding of the mechanisms involved in driving hepatic injury after these compounds perfuse and are metabolized by the liver. Herein we assess in vitro, in vivo (animal), and in silico strategies used to develop predictive DILI models. We address the effectiveness of several two- and three-dimensional in vitro cellular methods that are frequently employed in hepatotoxicity screens and how they can be used to predict DILI in humans. We also explore how humanized animal models can recapitulate human drug metabolic profiles and associated liver injury. Finally, we highlight the maturation of computational methods for predicting hepatotoxicity, the untapped potential of artificial intelligence for improving in silico DILI screens, and how knowledge acquired from these predictions can shape the refinement of experimental methods.

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.

vNN Web Server for ADMET Predictions

In drug development, early assessments of pharmacokinetic and toxic properties are important stepping stones to avoid costly and unnecessary failures. Considerable progress has recently been made in the development of computer-based (in silico) models to estimate such properties. Nonetheless, such models can be further improved in terms of their ability to make predictions more rapidly, easily, and with greater reliability. To address this issue, we have used our vNN method to develop 15 absorption, distribution, metabolism, excretion, and toxicity (ADMET) prediction models. These models quickly assess some of the most important properties of potential drug candidates, including their cytotoxicity, mutagenicity, cardiotoxicity, drug-drug interactions, microsomal stability, and likelihood of causing drug-induced liver injury. Here we summarize the ability of each of these models to predict such properties and discuss their overall performance. All of these ADMET models are publically available on our website (https://vnnadmet.bhsai.org/), which also offers the capability of using the vNN method to customize and build new models.

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.

CarcinoPred-EL: Novel models for predicting the carcinogenicity of chemicals using molecular fingerprints and ensemble learning methods

Carcinogenicity refers to a highly toxic end point of certain chemicals, and has become an important issue in the drug development process. In this study, three novel ensemble classification models, namely Ensemble SVM, Ensemble RF, and Ensemble XGBoost, were developed to predict carcinogenicity of chemicals using seven types of molecular fingerprints and three machine learning methods based on a dataset containing 1003 diverse compounds with rat carcinogenicity. Among these three models, Ensemble XGBoost is found to be the best, giving an average accuracy of 70.1 ± 2.9%, sensitivity of 67.0 ± 5.0%, and specificity of 73.1 ± 4.4% in five-fold cross-validation and an accuracy of 70.0%, sensitivity of 65.2%, and specificity of 76.5% in external validation. In comparison with some recent methods, the ensemble models outperform some machine learning-based approaches and yield equal accuracy and higher specificity but lower sensitivity than rule-based expert systems. It is also found that the ensemble models could be further improved if more data were available. As an application, the ensemble models are employed to discover potential carcinogens in the DrugBank database. The results indicate that the proposed models are helpful in predicting the carcinogenicity of chemicals. A web server called CarcinoPred-EL has been built for these models (http://ccsipb.lnu.edu.cn/toxicity/CarcinoPred-EL/).

Evolution of Experimental Models of the Liver to Predict Human Drug Hepatotoxicity and Efficacy

In this article, we review the past applications of in vitro models in identifying human hepatotoxins and then focus on the use of multiscale experimental models in drug development, including the use of zebrafish and human cell-based, 3-dimensional, microfluidic systems of liver functions as key components in applying Quantitative Systems Pharmacology (QSP). We have implemented QSP as a platform to improve the rate of success in the process of drug discovery and development of therapeutics.

Conformation-dependent QSAR approach for the prediction of inhibitory activity of bromodomain modulators

Epigenetic drug discovery is a promising research field with growing interest in the scientific community, as evidenced by the number of publications and the large amount of structure-epigenetic activity information currently available in the public domain. Computational methods are valuable tools to analyse and understand the activity of large compound collections from their structural information. In this manuscript, QSAR models to predict the inhibitory activity of a diverse and heterogeneous set of 88 organic molecules against the bromodomains BRD2, BRD3 and BRD4 are presented. A conformation-dependent representation of the chemical structures was established using the RDKit software and a training and test set division was performed. Several two-linear and three-linear QuBiLS-MIDAS molecular descriptors ( www.tomocomd.com ) were computed to extract the geometric structural features of the compounds studied. QuBiLS-MIDAS-based features sets, to be used in the modelling, were selected using dimensionality reduction strategies. The multiple linear regression procedure coupled with a genetic algorithm were employed to build the predictive models. Regression models containing between 6 to 9 variables were developed and assessed according to several internal and external validation methods. Analyses of outlier compounds and the applicability domain for each model were performed. As a result, the models against BRD2 and BRD3 with 8 variables and the model with 9 variables against BRD4 were those with the best overall performance according to the criteria accounted for. The results obtained suggest that the models proposed will be a good tool for studying the inhibitory activities of drug candidates against the bromodomains considered during epigenetic drug discovery.