Computational Toxicology最新文献

Artemisinin and its semi-synthetic derivatives are not only indicated for malaria but also cancer, inflammatory and autoimmune diseases. Its inflammatory and immunosuppressive target is PI3K/AKT pathways. The structural and kinetic aspect of the PI3K inhibition was investigated in the current study using computational approaches. Binding energies of dihydroartemisinin (DHA) to p¹¹⁰-PI3K-β was computed using the MMPBSA method in comparison with the standard inhibitor (GD9). Kinetic parameter (K_on/K_off) was also evaluated for the complexes using adaptive sampling protocols and Markov state model analysis. p¹¹⁰-PI3K- β dynamics and community network analysis were also performed following conventional Molecular dynamics simulation. The results showed −63.99 ± 1.53 and −74.14 ± 3.47 (Kj/mol) binding energies for DHA and GD9 respectively. K_on/K_off estimates for DHA and GD9 are 12.4, and 2.13 (M⁻¹) respectively. Analysis of the trajectories showed that DHA selectively partitions into p¹¹⁰-PI3K- β affinity pocket, forces open conformation, and kept catalytic pocket-M783 in a flat conformation whilst forcing large displacement around the C2-domain. In conclusion, DHA is a high affinity (slow-binding, slow-dissociating), flat-conformation p¹¹⁰-PI3K- β inhibitor.

The development and application of predictive models towards toxicity of engineered nanomaterials is still far from being satisfactory. One promising contribution to confront this challenge is to effectively augment the performance of machine learning classifiers by progressing the approach towards balancing experimental toxicity data. We propose an improved balancing methodology by fitting the in-vitro toxicological dose-response datasets of engineered nanomaterials to three, four, and five, free parameter dose-response models. The four-free parameter model displays the best fit (in terms of adjusted R²) for most of the examined data. The fitted curve yields, in each case, a continuous sequence of data points, which extends the restricted experimental data and generates additional fitted data points for the minority class, leading to the formation of balanced data for predicting the nanoparticle’s toxicology by decision tree classifiers. The ability to best predict the experimental toxicity data, by applying the decision tree model, was tested by forming three versions of the same experimental data: the imbalanced raw experimental data, the balanced data by applying the common Synthetic Minority Oversampling Technique, and by using the approach of Balanced Fitted Dose-Response method, introduced in the present study. We demonstrate that our approach provides improved performance of decision trees in predicting nanoparticles’ toxicity, a method that pertains also to chemical toxicity, central in health and environmental research.

As part of a wider toxicological risk assessment to ensure patient safety, extractables and leachables (E&Ls) which are observed above the relevant qualification threshold need to be assessed for their sensitisation potential. This study sought to investigate whether in silico toxicity models could be used to predict the sensitisation hazard and potency potential of E&Ls. An extensive dataset of relevant chemicals was collated by combining and standardising two lists of E&Ls previously published by ELSIE and the PQRI, resulting in a dataset of 790 unique materials. Sensitisation data was then located where possible, resulting in 290 chemicals being associated with dermal sensitisation hazard information, 106 chemicals with dermal sensitisation potency information, and 47 chemicals with respiratory sensitisation information. Existing expert knowledge, in the form of structural alerts within Derek Nexus, was able to accurately predict both the dermal and respiratory sensitisation potential of the E&Ls. 75 different statistical models were also built, using several algorithms and descriptors, and trained on the available dermal sensitisation data. A number of these models proved able to accurately predict the sensitisation potential of the E&Ls, which were found to occupy the same chemical space as the training sets. Finally, hybrid approaches combining expert knowledge and statistical models were investigated, including a tiered system where the skin sensitisation alerts in Derek Nexus provided a hazard prediction, followed by a potency prediction resulting from an alert-based k-nearest neighbours model. The inclusion of the Dermal Sensitisation Thresholds as default, worst-case scenario predictions in cases where similar chemicals were lacking ensured that a prediction was provided for every chemical. It is hoped that this novel workflow, which combines expert knowledge, a statistical model and existing toxicity thresholds, will aid toxicologists when assessing the sensitisation potential of E&Ls administered by any route of administration.

High-throughput screening (HTS) assays for bioactivity in the Tox21 program aim to evaluate an array of different biological targets and pathways, but a significant barrier to interpretation of these data is the lack of high-throughput screening (HTS) assays intended to identify non-specific reactive chemicals. This is an important aspect for prioritising chemicals to test in specific assays, identifying promiscuous chemicals based on their reactivity, as well as addressing hazards such as skin sensitisation which are not necessarily initiated by a receptor-mediated effect but act through a non-specific mechanism. Herein, a fluorescence-based HTS assay that allows the identification of thiol-reactive compounds was used to screen 7,872 unique chemicals in the Tox21 10 K chemical library. Active chemicals were compared with profiling outcomes using structural alerts encoding electrophilic information. Random Forest classification models based on chemical fingerprints were developed to predict assay outcomes and evaluated through 10-fold stratified cross validation (CV). The mean CV Balanced Accuracy of the validation set was 0.648. The model developed shows promise as a tool to screen untested chemicals for their potential electrophilic reactivity based solely on chemical structural features.

Ionic Liquids (ILs) are considered greener alternatives to traditional organic solvents due to their unique physical and chemical properties. Nevertheless, recent studies showed that ILs can induce toxic effects in ecosystem. Therefore, it is essential to determine the level of risk to the aquatic life to successfully use these ILs. Toxicity measurement of various ILs on a broad spectrum of conditions through experimental techniques is way demanding on time, resources, and is at times impractical. Various research works have been performed in Quantitative Property Relationship (QSAR/QSPR) for IL toxicity prediction expressed as EC50. In this study, five supervised machine learning models were trained and tested using nine Principal Properties (PPs) as descriptors to predict leukemia rat cell line (IPC-81) cytotoxicity. Then eight feature selection techniques were used to preprocess the data to improve the performance of the best machine learning model among the preliminary trained models. Analysis of the performance of the models on predicting the out-of-sample data set showed that the Extreme Gradient Boosting (XGBoost) supervised machine learning model is the best in predicting with the highest test score (R² = 0.79). This model was the most parsimonious (minimum AIC of 46.50), consistent (minimum RMSE of 0.45), and precise (minimum MAE of 0.32) in predicting IPC-81 cytotoxicity. The feature importance attribute of XGBoost confirmed that the structural features of ILs’ cation like cationic hydrophilicity and the side chain length have significant impact on the toxicity. Nevertheless, the anionic part of IL is also important to their toxicity and needs to be considered in toxicity prediction. Among the tested feature selection techniques, the random forest technique was the best in improving model performance (i.e., the least error matrices: AIC = 41.22, MAE = 0.31 and RMSE = 0.4259 respectively) but at longer execution time. However, the wrapper methods were the most robust in improving computational efficiency (i.e, improved the model performance at the shortest execution time). Therefore, this study improves QSPR studies on toxicity prediction of new ILs with the application of machine learning and feature selection techniques.

In early phases of the drug discovery process, evaluating the off-target pharmacology of a candidate drug is important when considering potential safety risks. Such off-target liabilities are most commonly evaluated using panels of in vitro pharmacology assays with strong association to well-defined toxicological events. In addition to in vitro panels, preliminary in silico evaluation is emerging as a valuable approach to support identification of potential off-target hits, even prior to synthesis of chemical material. To ascertain the utility of in silico target profiling, the predictive performance of a proprietary in silico predictive tool was evaluated against an in-house data set of 94 compounds with associated in vitro panel data, including binding inhibition and functional agonism/antagonism. Of the compounds tested, the primary target was predicted with 35% sensitivity. However, the sensitivity to predict the primary target decreased to 16% for a subset of compounds not reported within the Chemical Abstracts Service registry. For the known off-target hits for all tested compounds, the value of sensitivity was 16% for binding assays and 23% for functional assays. To better understand the applicability of the in silico off-target prediction, we performed in vitro binding assays, to evaluate five additional off-targets that were predicted by in silico but not covered by our standard off-target binding or functional panels. Although no new off-target hit was identified through this campaign, as technologies evolve, the in silico predictions could provide valuable insights to identify potential off-targets and mechanistic insights on target organ toxicities caused by compounds in in vivo studies.

In this study, a local quantitative structure–activity relationship (QSAR) models were developed for set of compounds tested for their inhibitory activity against six different cancer cell lines viz. A-549, MCF-7, HeLa, HepG-2, PaCa-2 and DLD-1. Two different statistical approaches Principal Component Analysis (PCA) and Partial Least Square (PLS) analyses were employed to developed QSAR models. Further, activity predictions were carried out for in-house synthesized 63 pyrazole derivatives. Prediction of pIC₅₀ value of all 63 synthesized pyrazole derivatives were estimated based on the most significant QSAR model developed for each cancer cell line. Several statistical parameters such as correlation coefficient R², RMSE, Cross validated R², Cross validated RMSE, internal validation Q² and the external validation R² revealed that developed models showed a significant value for explaining an acceptable QSAR model. The results derived highlighted some important compounds for being the most promise lead candidate against the six-cancer cell line with a significant pIC₅₀ value. Considering the contribution of most important descriptors, we have designed new molecules which found to have greater inhibitory potentiality than the reference compounds. Overall, the results suggest that the developed QSAR models might be useful as a theoretical reference for experimental studies and designing more potent anti-cancer therapeutic pyrazoles based compounds.

Chlorpyrifos is the toxicant chemical from the class of organophosphorus insecticides. The insecticide undergoes environmental degradation to chlorpyrifos‐oxon (CPYO), des‐ethyl chlorpyrifos (DEC), 3,5,6‐trichloro‐2‐methoxypyridine (TMP) and 3,5,6‐trichloro‐2‐pyridinol (TCP). Herein, CPF along with its degradants were optimized employing density functional theory (DFT) and B3LYP/6-311G+(d,p) basis set to elucidate their thermal and frontier molecular orbital properties. The DFT outcome revealed that TCP showed the lowest HOMO-LUMO gap (4.38 eV), also highest dipole moment, electrophilicity index and basicity. Docking was done using AutoDock 4.2.6 against human glutathione S-transferases to search binding affinity and interactions of all pollutants with the protein. The docking results expressed that TCP required least binding energy (−5.51 kcal mol⁻¹) which is relatable to the DFT studies and might act as the most powerful inhibitor. GROMACS 5.1.1 was utilized to perform simulation studies for each ligand–protein docked complexes. Results concluded that CPF, DEC, TMP, CPYO and TCP could possibly perform as toxic and inhibit enzymatic activity by interrupting the metabolic pathways in humans.

Chemoinformatics has been successfully employed in safety assessment through various regulatory programs for which information from databases, as well as predictive methodologies including computational methods, are accepted. One example is the European Union Cosmetics Products Regulations, for which Cosmetics Europe (CE) research activities in non-animal methods have been managed by the Long Range Science Strategy (LRSS) program. The vision is to use mechanistic aspects of existing non-animal methods, as well as New Approach Methodologies (NAMs), to demonstrate that safety assessment of chemicals can be performed using a combination of in silico and in vitro data. To this end, ChemTunes•ToxGPS® has been adopted as the foundation of the safety assessment system and provides a platform to integrate data and knowledge, and enable toxicity predictions and safety assessments, relevant to cosmetics industries. The ChemTunes•ToxGPS® platform provides chemical, biological, and safety data based both on experiments and predictions, and an interactive/customizable read-across platform. The safety assessment workflow enables users to compile qualified data sources, quantify their reliabilities, and combine them using a weight of evidence approach based on decision theory. The power of this platform was demonstrated through a use case to perform a safety assessment for Perilla frutescens through the workflows of threshold of toxicological concern (TTC), in silico predictions (QSAR and structural rules) and quantitative read-across (qRAX) assessment for overall safety. The system digitalizes workflows within a knowledge hub, exploiting advanced in silico tools in this age of artificial intelligence. The further design of the system for next generation risk assessment (NGRA) is scientifically guided by interactions between the workgroup and international regulatory entities.