hERG-toxicity prediction using traditional machine learning and advanced deep learning techniques.

The rise of artificial intelligence (AI) based algorithms has gained a lot of interest in the pharmaceutical development field. Our study demonstrates utilization of traditional machine learning techniques such as random forest (RF), support-vector machine (SVM), extreme gradient boosting (XGBoost), deep neural network (DNN) as well as advanced deep learning techniques like gated recurrent unit-based DNN (GRU-DNN) and graph neural network (GNN), towards predicting human ether-á-go-go related gene (hERG) derived toxicity. Using the largest hERG dataset derived to date, we have utilized 203,853 and 87,366 compounds for training and testing the models, respectively. The results show that GNN, SVM, XGBoost, DNN, RF, and GRU-DNN all performed well, with validation set AUC ROC scores equals 0.96, 0.95, 0.95, 0.94, 0.94 and 0.94, respectively. The GNN was found to be the top performing model based on predictive power and generalizability. The GNN technique is free of any feature engineering steps while having a minimal human intervention. The GNN approach may serve as a basis for comprehensive automation in predictive toxicology. We believe that the models presented here may serve as a promising tool, both for academic institutes as well as pharmaceutical industries, in predicting hERG-liability in new molecular structures.

In Silico Prediction of Eye Irritation Using Hansen Solubility Parameters and Predicted pKa Values.

An in silico method has been developed that permits the binary differentiation between pure liquids causing serious eye damage or eye irritation, and pure liquids with no need for such classification, according to the UN GHS system. The method is based on the finding that the Hansen Solubility Parameters (HSP) of a liquid are collectively important predictors for eye irritation. Thus, by applying a two-tier approach in which in silico-predicted pKa values (firstly) and a trained model based solely on in silico-predicted HSP data (secondly) were used, we have developed, and validated, a fully in silico approach for predicting the outcome of a Draize test (in terms of UN GHS Cat. 1/Cat. 2A/Cat. 2B or UN GHS No Cat.) with high validation set performance (sensitivity = 0.846, specificity = 0.818, balanced accuracy = 0.832) using SMILES only. The method is applicable to pure non-ionic liquids with molecular weight below 500 g/mol, fewer than six hydrogen bond donors (e.g. nitrogen-hydrogen or oxygen-hydrogen bonds) and fewer than eleven hydrogen bond acceptors (e.g. nitrogen or oxygen atoms). Due to its fully in silico characteristics, this method can be applied to pure liquids that are still at the desktop design stage and not yet in production.

Predicting Chemical-Induced Liver Toxicity Using High-Content Imaging Phenotypes and Chemical Descriptors: A Random Forest Approach.

Hepatotoxicity is a major reason for the withdrawal or discontinuation of drugs from clinical trials. Thus, better tools are needed to filter potential hepatotoxic drugs early in drug discovery. Our study demonstrates utilization of HCI phenotypes, chemical descriptors, and both combined (hybrid) descriptors to construct random forest classifiers (RFCs) for the prediction of hepatotoxicity. HCI data published by Broad Institute provided HCI phenotypes for about 30 000 samples in multiple replicates. Phenotypes belonging to 346 chemicals, which were tested in up to eight replicates, were chosen as a basis for our analysis. We then constructed individual RFC models for HCI phenotypes, chemical descriptors, and hybrid (chemical and HCI) descriptors. The model that was constructed using selective hybrid descriptors showed high predictive performance with 5-fold cross validation (CV) balanced accuracy (BA) at 0.71, whereas within the given applicability domain (AD), independent test set and external test set prediction BAs were equal to 0.61 and 0.60, respectively. The model constructed using chemical descriptors showed a similar predictive performance with a 5-fold CV BA equal to 0.66, a test set prediction BA within the AD equal to 0.56, and an external test set prediction BA within the AD equal to 0.50. In conclusion, the hybrid and chemical descriptor-based models presented here should be considered as a new tool for filtering hepatotoxic molecules during compound prioritization in drug discovery.

A k-nearest neighbor classification of hERG K(+) channel blockers.

A series of 172 molecular structures that block the hERG K(+) channel were used to develop a classification model where, initially, eight types of PaDEL fingerprints were used for k-nearest neighbor model development. A consensus model constructed using Extended-CDK, PubChem and Substructure count fingerprint-based models was found to be a robust predictor of hERG activity. This consensus model demonstrated sensitivity and specificity values of 0.78 and 0.61 for the internal dataset compounds and 0.63 and 0.54 for the external (PubChem) dataset compounds, respectively. This model has identified the highest number of true positives (i.e. 140) from the PubChem dataset so far, as compared to other published models, and can potentially serve as a basis for the prediction of hERG active compounds. Validating this model against FDA-withdrawn substances indicated that it may even be useful for differentiating between mechanisms underlying QT prolongation.

Acute Toxicity-Supported Chronic Toxicity Prediction: A k-Nearest Neighbor Coupled Read-Across Strategy.

A k-nearest neighbor (k-NN) classification model was constructed for 118 RDT NEDO (Repeated Dose Toxicity New Energy and industrial technology Development Organization; currently known as the Hazard Evaluation Support System (HESS)) database chemicals, employing two acute toxicity (LD50)-based classes as a response and using a series of eight PaDEL software-derived fingerprints as predictor variables. A model developed using Estate type fingerprints correctly predicted the LD50 classes for 70 of 94 training set chemicals and 19 of 24 test set chemicals. An individual category was formed for each of the chemicals by extracting its corresponding k-analogs that were identified by k-NN classification. These categories were used to perform the read-across study for prediction of the chronic toxicity, i.e., Lowest Observed Effect Levels (LOEL). We have successfully predicted the LOELs of 54 of 70 training set chemicals (77%) and 14 of 19 test set chemicals (74%) to within an order of magnitude from their experimental LOEL values. Given the success thus far, we conclude that if the k-NN model predicts LD50 classes correctly for a certain chemical, then the k-analogs of such a chemical can be successfully used for data gap filling for the LOEL. This model should support the in silico prediction of repeated dose toxicity.

Theoretical and Computational Strategies for the Study of the Molecular Imprinting Process and Polymer Performance.

The development of in silico strategies for the study of the molecular imprinting process and the properties of molecularly imprinted materials has been driven by a growing awareness of the inherent complexity of these systems and even by an increased awareness of the potential of these materials for use in a range of application areas. Here we highlight the development of theoretical and computational strategies that are contributing to an improved understanding of the mechanisms underlying molecularly imprinted material synthesis and performance, and even their rational design.

Towards global QSAR model building for acute toxicity: Munro database case study.

A series of 436 Munro database chemicals were studied with respect to their corresponding experimental LD50 values to investigate the possibility of establishing a global QSAR model for acute toxicity. Dragon molecular descriptors were used for the QSAR model development and genetic algorithms were used to select descriptors better correlated with toxicity data. Toxic values were discretized in a qualitative class on the basis of the Globally Harmonized Scheme: the 436 chemicals were divided into 3 classes based on their experimental LD50 values: highly toxic, intermediate toxic and low to non-toxic. The k-nearest neighbor (k-NN) classification method was calibrated on 25 molecular descriptors and gave a non-error rate (NER) equal to 0.66 and 0.57 for internal and external prediction sets, respectively. Even if the classification performances are not optimal, the subsequent analysis of the selected descriptors and their relationship with toxicity levels constitute a step towards the development of a global QSAR model for acute toxicity.

Binding site characterization of G protein-coupled receptor by alanine-scanning mutagenesis using molecular dynamics and binding free energy approach: application to C-C chemokine receptor-2 (CCR2).

The C-C chemokine receptor 2 (CCR2) was proved as a multidrug target in many diseases like diabetes, inflammation and AIDS, but rational drug design on this target is still lagging behind as the information on the exact binding site and the crystal structure is not yet available. Therefore, for a successful structure-based drug design, an accurate receptor model in ligand-bound state is necessary. In this study, binding-site residues of CCR2 was determined using in silico alanine scanning mutagenesis and the interactions between TAK-779 and the developed homology model of CCR2. Molecular dynamic simulation and Molecular Mechanics-Generalized Born Solvent Area method was applied to calculate binding free energy difference between the template and mutated protein. Upon mutating 29 amino acids of template protein and comparison of binding free energy with wild type, six residues were identified as putative hot spots of CCR2.

Molecular dynamics directed CoMFA studies on carbocyclic neuraminidase inhibitors.

Zanamivir is the known potent anti-influenza agent targeting the key enzyme neuraminidase that cleaves sialic acid from cell receptors allowing release of newly formed virions. Molecular dynamics simulation was carried out to determine the dynamic behavior of Zanamivir upon its binding to flexible loops of neuraminidase and to analyse its interactions in the bioactive state. Neuraminidase exhibits wide range of affinity with structurally similar compounds. CoMFA study was used to determine quantitative structure-activity relationship for 36 carbocyclic Neuraminidase inhibitors (NIs). The CoMFA model was also successfully built using cross-validated r²cv = 0.580 and r²pred = 0.638.

Rational design of CCR2 antagonists: a survey of computational studies.

IMPORTANCE OF THE FIELD: CC-chemokine receptor 2 (CCR2) belongs to the GPCR superfamily and is the primary receptor for monocyte chemoattractant protein-1 (MCP-1), also known as chemokine ligand CCL2. Studies indicate the possible involvement of MCP-1 and CCR2 in various disease conditions, such as rheumatic arthritis, multiple sclerosis, vascular diseases, obesity and diabetes, via the inflammatory pathway. MCP-1 and CCR2 knockout mice under a broad range of stimuli exhibit deficient monocyte recruitment suggesting its potential role in inflammation. Overall, there is evidence that an impairment of monocyte trafficking in inflammation models occurs when there is a loss of MCP-1 effector function. This makes its receptor, CCR2, an attractive target for pharmaceutical research. Several small molecular CCR2 antagonists have been developed, particularly in the industry. AREAS COVERED IN THIS REVIEW: In this article, we have summarized the in silico work carried out in the area of CCR2 and reviewed mainly the computer aided drug design (CADD) studies reported on quantitative structure-activity relationship, homology modeling, molecular docking and virtual screening. WHAT THE READER WILL GAIN: A survey of computational studies for the rational design and development of CCR2 antagonists. TAKE HOME MESSAGE: CADD tools can be used to rationalize the identification of the potential leads and these techniques can be effectively applied in the rapid searching of novel and potent CCR2 antagonists.