Machine-learning based prediction models for assessing skin irritation and corrosion potential of liquid chemicals using physicochemical properties by XGBoost

Skin irritation test is an essential part of the safety assessment of chemicals. Recently, computational models to predict the skin irritation draw attention as alternatives to animal testing. We developed prediction models on skin irritation/corrosion of liquid chemicals using machine learning algorithms, with 34 physicochemical descriptors calculated from the structure. The training and test dataset of 545 liquid chemicals with reliable in vivo skin hazard classifications based on UN Globally Harmonized System [category 1 (corrosive, Cat 1), 2 (irritant, Cat 2), 3 (mild irritant, Cat 3), and no category (nonirritant, NC)] were collected from public databases. After the curation of input data through removal and correlation analysis, every model was constructed to predict skin hazard classification for liquid chemicals with 22 physicochemical descriptors. Seven machine learning algorithms [Logistic regression, Naïve Bayes, k-nearest neighbor, Support vector machine, Random Forest, Extreme gradient boosting (XGB), and Neural net] were applied to ternary and binary classification of skin hazard. XGB model demonstrated the highest accuracy (0.73-0.81), sensitivity (0.71-0.92), and positive predictive value (0.65-0.81). The contribution of physicochemical descriptors to the classification was analyzed using Shapley Additive exPlanations plot to provide an insight into the skin irritation of chemicals.

Supplementary Information

The online version contains supplementary material available at 10.1007/s43188-022-00168-8.

In-Silico Drug Toxicity and Interaction Prediction for Plant Complexes Based on Virtual Screening and Text Mining

Potential drug toxicities and drug interactions of redundant compounds of plant complexes may cause unexpected clinical responses or even severe adverse events. On the other hand, super-additivity of drug interactions between natural products and synthetic drugs may be utilized to gain better performance in disease management. Although without enough datasets for prediction model training, based on the SwissSimilarity and PubChem platforms, for the first time, a feasible workflow of prediction of both toxicity and drug interaction of plant complexes was built in this study. The optimal similarity score threshold for toxicity prediction of this system is 0.6171, based on an analysis of 20 different herbal medicines. From the PubChem database, 31 different sections of toxicity information such as "Acute Effects", "NIOSH Toxicity Data", "Interactions", "Hepatotoxicity", "Carcinogenicity", "Symptoms", and "Human Toxicity Values" sections have been retrieved, with dozens of active compounds predicted to exert potential toxicities. In Spatholobus suberectus Dunn (SSD), there are 9 out of 24 active compounds predicted to play synergistic effects on cancer management with various drugs or factors. The synergism between SSD, luteolin and docetaxel in the management of triple-negative breast cancer was proved by the combination index assay, synergy score detection assay, and xenograft model.

Comparing Multiple Machine Learning Algorithms and Metrics for Estrogen Receptor Binding Prediction

Many chemicals that disrupt endocrine function have been linked to a variety of adverse biological outcomes. However, screening for endocrine disruption using in vitro or in vivo approaches is costly and time-consuming. Computational methods, e.g., quantitative structure-activity relationship models, have become more reliable due to bigger training sets, increased computing power, and advanced machine learning algorithms, such as multilayered artificial neural networks. Machine learning models can be used to predict compounds for endocrine disrupting capabilities, such as binding to the estrogen receptor (ER), and allow for prioritization and further testing. In this work, an exhaustive comparison of multiple machine learning algorithms, chemical spaces, and evaluation metrics for ER binding was performed on public data sets curated using in-house cheminformatics software (Assay Central). Chemical features utilized in modeling consisted of binary fingerprints (ECFP6, FCFP6, ToxPrint, or MACCS keys) and continuous molecular descriptors from RDKit. Each feature set was subjected to classic machine learning algorithms (Bernoulli Naive Bayes, AdaBoost Decision Tree, Random Forest, Support Vector Machine) and Deep Neural Networks (DNN). Models were evaluated using a variety of metrics: recall, precision, F1-score, accuracy, area under the receiver operating characteristic curve, Cohen's Kappa, and Matthews correlation coefficient. For predicting compounds within the training set, DNN has an accuracy higher than that of other methods; however, in 5-fold cross validation and external test set predictions, DNN and most classic machine learning models perform similarly regardless of the data set or molecular descriptors used. We have also used the rank normalized scores as a performance-criteria for each machine learning method, and Random Forest performed best on the validation set when ranked by metric or by data sets. These results suggest classic machine learning algorithms may be sufficient to develop high quality predictive models of ER activity.

De Novo Design of Ligands Using Computational Methods

De novo design technique is complementary to high-throughput virtual screening and is believed to contribute in pharmaceutical development of novel drugs with desired properties at a very low cost and time-efficient manner. In this chapter, we outline the basic de novo design concepts based on computational methods with an example.

Evaluation of existing (Q)SAR models for skin and eye irritation and corrosion to use for REACH registration

The performance of the (Q)SAR models Derek Nexus, Toxtree and Case Ultra for the prediction of skin and eye irritation/corrosion is investigated. For irritation and corrosion of the skin, 117 compounds and for the eye, 125 compounds were listed. The balance between the groups positive and negative for irritation and corrosion was maintained. The obtained predictions were compared with experimental data and the numbers of true and false positives and negatives were determined. Based on these results several performance parameters of the tested (Q)SAR models were calculated. Despite all the efforts to make good and valid models, the results indicate a poor predictivity of the current models: a lot of compounds were not predicted, were out of the applicability domain or were predicted wrong. Considering our results, it can be concluded that the tested models are not yet sufficiently powerful for implementation. Possibly the training-sets used within the current models are not yet comprehensive enough or the incorporated data are not of enough quality. Although the use of these models as stand-alone evaluation is not recommended, these models can be of value as weight-of-evidence in the context of expert knowledge in an Integrated Approach to Testing and Assessment.

Large-Scale Predictive Drug Safety: From Structural Alerts to Biological Mechanisms

The recent explosion of data linking drugs, proteins, and pathways with safety events has promoted the development of integrative systems approaches to large-scale predictive drug safety. The added value of such approaches is that, beyond the traditional identification of potentially labile chemical fragments for selected toxicity end points, they have the potential to provide mechanistic insights for a much larger and diverse set of safety events in a statistically sound nonsupervised manner, based on the similarity to drug classes, the interaction with secondary targets, and the interference with biological pathways. The combined identification of chemical and biological hazards enhances our ability to assess the safety risk of bioactive small molecules with higher confidence than that using structural alerts only. We are still a very long way from reliably predicting drug safety, but advances toward gaining a better understanding of the mechanisms leading to adverse outcomes represent a step forward in this direction.

Investigations on the interactions of λphage-derived peptides against the SrtA mechanism in Bacillus anthracis

Bacillus anthracis is a well-known bioweapon pathogen, which coordinates the expression of its virulence factors in response to a specific environmental signal by its protein architecture. Absences of sortase signal functioning may fail to assemble the surface linked proteins and so B. anthracis cannot sustain an infection with host cells. Targeting the signaling mechanism of B. anthracis can be achieved by inhibition of SrtA enzyme through λphage-derived plyG. The lysin enzyme plyG is experimentally proven as bacteriolytic agent, specifically kill's B. anthracis by inhibiting the SrtA. Here, we have screened the peptides from λphage lysin, and these peptides are having the ability as LPXTG competitive inhibitors. In comparison to the activator peptide LPXTG binding motif, λphage lysin based inhibitor peptides are having much supremacy towards binding of SrtA. Finally, peptide structures extracted from PlyG are free from toxic, allergic abilities and also have the ability to terminate the signal transduction mechanism in B. anthracis.

Expert Systems for Toxicity Prediction

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

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