Estimating sensory irritation potency of volatile organic chemicals using QSARs based on decision tree methods for regulatory purpose

Establishing short-term occupational exposure limits (STELs) for sensory irritants using predictive and in silico respiratory rate depression (RD50) models

Sensory irritation is a health endpoint that serves as the critical effect basis for many occupational exposure limits (OELs). Schaper 1993 described a significant relationship with high correlation between the measured exposure concentration producing a 50% respiratory rate decrease (RD50) in a standard rodent assay and the American Conference of Governmental Industrial Hygienists (ACGIH®) Threshold Limit Values (TLVs®) as time-weighted averages (TWAs) for airborne chemical irritants. The results demonstrated the potential use of the RD50 values for deriving full-shift TWA OELs protective of irritant responses. However, there remains a need to develop a similar predictive model for deriving workplace short-term exposure limits (STELs) for sensory irritants. The aim of our study was to establish a model capable of correlating the relationship between RD50 values and published STELs to prospectively derive short-term exposure OELs for sensory irritants. A National Toxicology Program (NTP) database that included chemicals with both an RD50 and established STELs was used to fit several linear regression models. A strong correlation between RD50s and STELs was identified, with a predictive equation of ln (STEL) (ppm) = 0.86 * ln (RD50) (ppm) - 2.42 and an R2 value of 0.75. This model supports the use of RD50s to derive STELs for chemicals without existing exposure recommendations. Further, for data-poor sensory irritants, predicted RD50 values from in silico quantitative structure activity relationship (QSAR) models can be used to derive STELs. Hence, in silico methods and statistical modeling can present a path forward for establishing reliable OELs and improving worker safety and health.

Novel Models for Accurate Estimation of Air-Blood Partitioning: Applications to Individual Compounds and Complex Mixtures of Neutral Organic Compounds

The air-blood partition coefficient (Kab) is extensively employed in human health risk assessment for chemical exposure. However, current Kab estimation approaches either require an extensive number of parameters or lack precision. In this study, we present two novel and parsimonious models to accurately estimate Kab values for individual neutral organic compounds, as well as their complex mixtures. The first model, termed the GC×GC model, was developed based on the retention times of nonpolar chemical analytes on comprehensive two-dimensional gas chromatography (GC×GC). This model is unique in its ability to estimate the Kab values for complex mixtures of nonpolar organic chemicals. The GC×GC model successfully accounted for the Kab variance (R2 = 0.97) and demonstrated strong prediction power (RMSE = 0.31 log unit) for an independent set of nonpolar chemical analytes. Overall, the GC×GC model can be used to estimate Kab values for complex mixtures of neutral organic compounds. The second model, termed the partition model (PM), is based on two types of partition coefficients: octanol to water (Kow) and air to water (Kaw). The PM was able to effectively account for the variability in Kab data (n = 344), yielding an R2 value of 0.93 and root-mean-square error (RMSE) of 0.34 log unit. The predictive power and explanatory performance of the PM were found to be comparable to those of the parameter-intensive Abraham solvation models (ASMs). Additionally, the PM can be integrated into the software EPI Suite, which is widely used in chemical risk assessment for initial screening. The PM provides quick and reliable estimation of Kab compared to ASMs, while the GC×GC model is uniquely suited for estimating Kab values for complex mixtures of neutral organic compounds. In summary, our study introduces two novel and parsimonious models for the accurate estimation of Kab values for both individual compounds and complex mixtures.

Reliable prediction of sensory irritation threshold values of organic compounds using new models based on linear free energy relationships and GC×GC retention parameters

The human sensory irritation threshold (SIT) is an important biochemical parameter for the exposure assessment of organic air pollutants. First, we recalibrated the Abraham solvation models (ASMs) for 9 SIT endpoints by curating 720 individual experimental SIT values to find an accurate and parsimonious ASM variant, which exhibited root mean square error (RMSE) = 0.174-0.473 log unit. Second, we report linear free energy relationships - henceforth called partition models (PMs) - which exploit the correlations of 9 SIT endpoints with the linear combinations of partition coefficients for octanol-water and air-water systems showing RMSE = 0.221-0.591 log unit. These PMs can easily be integrated into widely used EPI-Suite™ screening tool. The explanatory and predictive performance of PMs were like parameter-intensive ASMs. Third, we present GC × GC models that are based on the retention times of the nonpolar analytes on the comprehensive two-dimensional gas chromatography (GC × GC), which successfully described the SIT variance (R²=0.959-0.996) and depicted a strong predictive power (RMSE = 0.359-0.660 log unit) for an independent set of nonpolar analytes. Taken together, PMs allow easy SIT screening of organic chemicals compared to ASMs. Unlike ASMs, our GC × GC models can be applied to estimate SIT of complex nonpolar mixtures.

RespiraTox - Development of a QSAR model to predict human respiratory irritants

Respiratory irritation is an important human health endpoint in chemical risk assessment. There are two established modes of action of respiratory irritation, 1) sensory irritation mediated by the interaction with sensory neurons, potentially stimulating trigeminal nerve, and 2) direct tissue irritation. The aim of our research was to, develop a QSAR method to predict human respiratory irritants, and to potentially reduce the reliance on animal testing for the identification of respiratory irritants. Compounds are classified as irritating based on combined evidence from different types of toxicological data, including inhalation studies with acute and repeated exposure. The curated project database comprised 1997 organic substances, 1553 being classified as irritating and 444 as non-irritating. A comparison of machine learning approaches, including Logistic Regression (LR), Random Forests (RFs), and Gradient Boosted Decision Trees (GBTs), showed, the best classification was obtained by GBTs. The LR model resulted in an area under the curve (AUC) of 0.65, while the optimal performance for both RFs and GBTs gives an AUC of 0.71. In addition to the classification and the information on the applicability domain, the web-based tool provides a list of structurally similar analogues together with their experimental data to facilitate expert review for read-across purposes.

In silico approaches in organ toxicity hazard assessment: Current status and future needs for predicting heart, kidney and lung toxicities

The kidneys, heart and lungs are vital organ systems evaluated as part of acute or chronic toxicity assessments. New methodologies are being developed to predict these adverse effects based on in vitro and in silico approaches. This paper reviews the current state of the art in predicting these organ toxicities. It outlines the biological basis, processes and endpoints for kidney toxicity, pulmonary toxicity, respiratory irritation and sensitization as well as functional and structural cardiac toxicities. The review also covers current experimental approaches, including off-target panels from secondary pharmacology batteries. Current in silico approaches for prediction of these effects and mechanisms are described as well as obstacles to the use of in silico methods. Ultimately, a commonly accepted protocol for performing such assessment would be a valuable resource to expand the use of such approaches across different regulatory and industrial applications. However, a number of factors impede their widespread deployment including a lack of a comprehensive mechanistic understanding, limited in vitro testing approaches and limited in vivo databases suitable for modeling, a limited understanding of how to incorporate absorption, distribution, metabolism, and excretion (ADME) considerations into the overall process, a lack of in silico models designed to predict a safe dose and an accepted framework for organizing the key characteristics of these organ toxicants.

A simple approach to the prediction of soil sorption of organophosphorus pesticides

Organophosphorus pesticides (OP) affect the crops and environments, and the reliable approach to the prediction of soil sorption of pesticides is required. In this respect, we proposed a simple Chemometrics approach, in which the Tchebichef image moment (TM) method was used to extract useful information from the greyscale images of molecular structures and the quantitative model was established by stepwise regression to predict the soil sorption of OPs. Different squared correlation coefficients including the leave-one-out cross-validation (LOO-CV) (Q²) that concerns the training set and the (R²_test) which concerns the external independent test set are more than 0.96. This reflects that the established model has considerably high accuracy and reliability. Compared with the literature on the strategies of quantitative structure-property relationship (QSPR), the proposed method is more suitable, in which the established model shows a high predictive ability. Our study provides another effective approach to predict the soil sorption of OPs and also extends the innovative pathway of QSPR modelling.

Beyond dermal exposure: The respiratory tract as a target organ in hazard assessments of cosmetic ingredients

Dermal contact is the main route of exposure for most cosmetics; however, inhalation exposure could be significant for some formulations (e.g., aerosols, powders). Current cosmetic regulations do not require specific tests addressing respiratory irritation and sensitisation, and despite the prohibition of animal testing for cosmetics, no alternative methods have been validated to assess these endpoints to date. Inhalation hazard is mainly determined based on existing human and animal evidence, read-across, and extrapolation of data from different target organs or tissues, such as the skin. However, because of mechanistic differences, effects on the skin cannot predict effects on the respiratory tract, which indicates a substantial need for the development of new approach methodologies addressing respiratory endpoints for inhalable chemicals in general. Cosmetics might present a particularly significant need for risk assessments of inhalation exposure to provide a more accurate toxicological evaluation and ensure consumer safety. This review describes the differences in the mechanisms of irritation and sensitisation between the skin and the respiratory tract, the progress that has already been made, and what still needs to be done to fill the gap in the inhalation risk assessment of cosmetic ingredients.

Versatile in silico modeling of partition coefficients of organic compounds in polydimethylsiloxane using linear and nonlinear methods

Environmental fate, behavior and effects of hazardous organic compounds have recently received great attention in diverse environmental phases, including water, atmosphere, soil and sediment. Considering polydimethylsiloxane (PDMS) fibers were validated for the wide application in the determination of partition behavior in passive sampling, in this work, several in silico models were established to predict PDMS-water (K_PDMS-w), PDMS-air (K_PDMS-a) and PDMS-seawater partition coefficients (K_PDMS-sw) of diverse chemicals. This is an attempt to combine conventional linear method and popular nonlinear algorithm for the estimation of partition coefficients between PDMS and different environmental media. All of the developed models showed satisfactory goodness-of-fit with high adjusted correlation coefficient (R²_adj) and were validated to be robust, stable and predictable by various internal and external validation techniques, deriving a wide series of statistical checks. Moreover, it was found that hydrophobicity, polarizability, charge distribution and molecular size of compounds contributed significantly to the model development by interpreting the selected descriptors. Based on the broad applicability domains (ADs), the current study provides suitable tools to fill the experimental data gap for other compounds and to help researchers better understand the mechanistic basis of adsorption behavior of PDMS.

Quantitative structure activity relationship (QSAR) modelling of the degradability rate constant of volatile organic compounds (VOCs) by OH radicals in atmosphere

The reaction with hydroxyl radicals (•OH) is an important way to remove the most volatile organic compounds (VOCs) in atmospheric environment. Thus, the reaction rate constant (k_OH) is important for assessing the persistence and exposure risk of VOCs, and is of great significance in evaluating the ecological risk of volatile organic chemicals. Fukui indices and bond order have a large effect on the degradation of VOCs, but so far, quantitative structure activity relationship (QSAR) models for VOCs degradation have rarely been considered these two factors. In this study, these two momentous factors will be considered along with other relevant quantitative parameters. A total of 180 substances are divided into training set (144 substances) and test set (36 substances), which are used to build and validate quantitative structure activity relationship (QSAR) models, respectively. Internal, external verification and y-randomization tests showed that the established model had excellent stability and reliability. The energy of the highest occupied molecular orbital (E_HOMO), the possibility of being attacked by radicals (f (0)_n) and the breaking of chemical bonds (BO_x) are the main factors affecting VOCs removal. Finally, the scope of the application domain was determined and the robustness of the model was further verified.

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.

From machine learning to deep learning: progress in machine intelligence for rational drug discovery

Machine intelligence, which is normally presented as artificial intelligence, refers to the intelligence exhibited by computers. In the history of rational drug discovery, various machine intelligence approaches have been applied to guide traditional experiments, which are expensive and time-consuming. Over the past several decades, machine-learning tools, such as quantitative structure-activity relationship (QSAR) modeling, were developed that can identify potential biological active molecules from millions of candidate compounds quickly and cheaply. However, when drug discovery moved into the era of 'big' data, machine learning approaches evolved into deep learning approaches, which are a more powerful and efficient way to deal with the massive amounts of data generated from modern drug discovery approaches. Here, we summarize the history of machine learning and provide insight into recently developed deep learning approaches and their applications in rational drug discovery. We suggest that this evolution of machine intelligence now provides a guide for early-stage drug design and discovery in the current big data era.

Room-temperature and temperature-dependent QSRR modelling for predicting the nitrate radical reaction rate constants of organic chemicals using ensemble learning methods

Experimental determinations of the rate constants of the reaction of NO3 with a large number of organic chemicals are tedious, and time and resource intensive; and the development of computational methods has widely been advocated. In this study, we have developed room-temperature (298 K) and temperature-dependent quantitative structure-reactivity relationship (QSRR) models based on the ensemble learning approaches (decision tree forest (DTF) and decision treeboost (DTB)) for predicting the rate constant of the reaction of NO3 radicals with diverse organic chemicals, under OECD guidelines. Predictive powers of the developed models were established in terms of statistical coefficients. In the test phase, the QSRR models yielded a correlation (r(2)) of >0.94 between experimental and predicted rate constants. The applicability domains of the constructed models were determined. An attempt has been made to provide the mechanistic interpretation of the selected features for QSRR development. The proposed QSRR models outperformed the previous reports, and the temperature-dependent models offered a much wider applicability domain. This is the first report presenting a temperature-dependent QSRR model for predicting the nitrate radical reaction rate constant at different temperatures. The proposed models can be useful tools in predicting the reactivities of chemicals towards NO3 radicals in the atmosphere, hence, their persistence and exposure risk assessment.

Modeling the toxicity of chemical pesticides in multiple test species using local and global QSTR approaches

The safety assessment processes require the toxicity data of chemicals in multiple test species and thus, emphasize the need for computational methods capable of toxicity prediction in multiple test species. Pesticides are designed toxic substances and find extensive applications worldwide. In this study, we have established local and global QSTR (quantitative structure-toxicity relationship) and ISC QSAAR (interspecies correlation quantitative structure activity-activity relationship) models for predicting the toxicities of pesticides in multiple aquatic test species using the toxicity data in crustacean (Daphnia magna, Americamysis bahia, Gammarus fasciatus, and Penaeus duorarum) and fish (Oncorhynchus mykiss and Lepomis macrochirus) species in accordance with the OECD guidelines. The ensemble learning based QSTR models (decision tree forest, DTF and decision tree boost, DTB) were constructed and validated using several statistical coefficients derived on the test data. In all the QSTR and QSAAR models, Log P was an important predictor. The constructed local, global and interspecies QSAAR models yielded high correlations (R2) of >0.941; >0.943 and >0.826, respectively between the measured and model predicted endpoint toxicity values in the test data. The performances of the local and global QSTR models were comparable. Furthermore, the chemical applicability domains of these QSTR/QSAAR models were determined using the leverage and standardization approaches. The results suggest for the appropriateness of the developed QSTR/QSAAR models to reliably predict the aquatic toxicity of structurally diverse pesticides in multiple test species and can be used for the screening and prioritization of new pesticides.