Graph neural networks-enhanced relation prediction for ecotoxicology (GRAPE)

Exposure to toxic chemicals threatens species and ecosystems. This study introduces a novel approach using Graph Neural Networks (GNNs) to integrate aquatic toxicity data, providing an alternative to complement traditional in vivo ecotoxicity testing. This study pioneers the application of GNN in ecotoxicology by formulating the problem as a relation prediction task. GRAPE's key innovation lies in simultaneously modelling 444 aquatic species and 2826 chemicals within a graph, leveraging relations from existing datasets where informative species and chemical features are augmented to make informed predictions. Extensive evaluations demonstrate the superiority of GRAPE over Logistic Regression (LR) and Multi-Layer Perceptron (MLP) models, achieving remarkable improvements of up to a 30% increase in recall values. GRAPE consistently outperforms LR and MLP in predicting novel chemicals and new species. In particular, GRAPE showcases substantial enhancements in recall values, with improvements of ≥ 100% for novel chemicals and up to 13% for new species. Specifically, GRAPE correctly predicts the effects of novel chemicals (104 out of 126) and effects on new species (7 out of 8). Moreover, the study highlights the effectiveness of the proposed chemical features and induced network topology through GNN for accurately predicting metallic (74 out of 86) and organic (612 out of 674) chemicals, showcasing the broad applicability and robustness of the GRAPE model in ecotoxicological investigations. The code/data are provided at https://github.com/csiro-robotics/GRAPE.

QSAR study of indole derivatives as active agents against Candida albicans: a DFT calculation

Indole and its derivatives are common heterocyclic compounds in nature that have a wider range of medicinal activities such as antifungal, anti-inflammatory, and anti-seizure. Virtually all indole derivatives showed outstanding antifungal activity against Candida albicans. The aim of this study was to QSAR modeling of indole derivatives and the design of new drugs that have antifungal activity. In this study, 52 compounds were selected. All optimized compounds and quantum descriptors were obtained using Gaussian software and DFT/B3LYP computational method with 6-31 G (d) basis set al, so other descriptors were determined using Dragon software. To examine the relationship between these descriptors and the activity of these compounds, the MLR linear correlation method was used, and the QSAR equation with R² = 0.7884 and R = 0.8879 was obtained for it. Likewise, MSE = 0.1897, RMSE = 0.2848, and Q² = 0.68663 approve the acceptability of the obtained model. The obtained equation reveals that the activity of these compounds is related to the negative coefficient of GATS8p, R7e +, and G2e, which means that with increasing the values ​​of these description nodes, the amount of activity declines. On the other hand, the activity of these compounds depended on the positive coefficients of HATS3p, MATS5e, and RDF045, i.e. with increasing these values, the activity of these compounds also increases, and a good correlation was obtained between the experimental and predicted activity values.

Predicting protection capacities of pyrimidine-based corrosion inhibitors for mild steel/HCl interface using linear and nonlinear QSPR models

Pyrimidine compounds have proven to be effective and efficient additives capable of protecting mild steel in acidic media. This class of organic compounds often functions as adsorption-type inhibitors of corrosion by forming a protective layer on the metallic substrate. The present study reports a computational study of forty pyrimidine compounds that have been investigated as sustainable inhibitors of mild steel corrosion in molar HCl solution. Quantitative structure property relationship was conducted using linear (multiple linear regression) and nonlinear (artificial neural network) models. Standardization method was employed in variable selection yielding five top chemical descriptors utilized for model development along with the inhibitor concentration. Multiple linear regression model yielded a fair predictive model. Artificial neural network model developed using k-fold cross-validation method provided a comprehensive insight into the corrosion protection mechanism of studied pyrimidine-based corrosion inhibitors. Using a multilayer perceptron with Levenberg-Marquardt algorithm, the study obtained the optimal model having a MSE of 8.479, RMSE of 2.912, MAD of 1.791, and MAPE of 2.648. The optimal neural network model was further utilized to forecast the protection capacities of nine non-synthesized pyrimidine derivatives. The predicted inhibition efficiencies ranged from 89 to 98%, revealing the significance of the considered chemical descriptors, the predictive capacity of the developed model, and the potency of the theoretical inhibitors.

A novel model for prediction of stability constants of the thiosemicarbazone ligands with different types of toxic heavy metal ions using structural parameters and multivariate linear regression method

A novel model is presented for reliable estimation of the stability constants of the thiosemicarbazone ligands with different types of toxic heavy metal ions (log β₁₁) in an aqueous solution, which has wide usage in environmental safety and ecotoxicology applications. The biggest reported data of log β₁₁ for 120 metalthiosemicarbazone complexes are used for deriving and testing the novel model. In contrast to available methods where they need the two-dimensional (2D) and three-dimensional (3D) complex molecular descriptors as well as expert users and computer codes, the novel correlation uses four additive and two non-additive structural parameters of thiosemicarbazone ligands. The calculated results of the novel correlation are compared with the outputs of the genetic algorithm with multivariate linear regression method (GA-MLR) as one of the best existing methods, which requires seven complex descriptors. The estimated results for 78 of training as well as 42 of two different test sets were established by external and internal validations. The values of statistical parameters comprising average deviation, average absolute deviation, average absolute relative deviation, absolute maximum deviation, and the coefficient of determination for 73 data of training set of New model/GA-MLR are 0.04/ - 0.25, 1.06/1.31, 14.4/18.7, 3.18/7.92, and 0.830/0.652, respectively. Thus, the predicted results of the new model are worthy as compared to the complex GA-MLR model. Moreover, assessments of various statistical parameters confirm that the new model provides great reliability, goodness-of-fit, accuracy, and precision.

Computer Aided Drug Design Methodologies with Natural Products in the Drug Research Against Alzheimer's Disease

Natural products are compounds isolated from plants that provide a variety of lead structures for the development of new drugs by the pharmaceutical industry. The interest in these substances increases because of their beneficial effects on human health. Alzheimer's disease (AD) affects occur in about 80% of individuals aged 65 years. AD, the most common cause of dementia in elderly people, is characterized by progressive neurodegenerative alterations, as decrease of cholinergic impulse, increased toxic effects caused by reactive oxygen species and the inflammatory process that the amyloid plaque participates. In silico studies is relevant in the process of drug discovery; through technological advances in the areas of structural characterization of molecules, computational science and molecular biology have contributed to the planning of new drugs used against neurodegenerative diseases. Considering the social impairment caused by an increased incidence of disease and that there is no chemotherapy treatment effective against AD; several compounds are studied. In the researches for effective neuroprotectants as potential treatments for Alzheimer's disease, natural products have been extensively studied in various AD models. This study aims to carry out a literature review with articles that address the in silico studies of natural products aimed at potential drugs against Alzheimer's disease (AD) in the period from 2015 to 2021.

Molecular modelling of compounds used for corrosion inhibition studies: a review

Molecular modelling of organic compounds using computational software has emerged as a powerful approach for theoretical determination of the corrosion inhibition potential of organic compounds. Some of the common techniques involved in the theoretical studies of corrosion inhibition potential and mechanisms include density functional theory (DFT), molecular dynamics (MD) and Monte Carlo (MC) simulations, and artificial neural network (ANN) and quantitative structure-activity relationship (QSAR) modeling. Using computational modelling, the chemical reactivity and corrosion inhibition activities of organic compounds can be explained. The modelling can be regarded as a time-saving and eco-friendly approach for screening organic compounds for corrosion inhibition potential before their wet laboratory synthesis would be carried out. Another advantage of computational modelling is that molecular sites responsible for interactions with metallic surfaces (active sites or adsorption sites) and the orientation of organic compounds can be easily predicted. Using different theoretical descriptors/parameters, the inhibition effectiveness and nature of the metal-inhibitor interactions can also be predicted. The present review article is a collection of major advancements in the field of computational modelling for the design and testing of the corrosion inhibition effectiveness of organic corrosion inhibitors.

Effects of Phthalate Esters (PAEs) on Cell Viability and Nrf2 of HepG2 and 3D-QSAR Studies

Phthalate esters (PAEs) are a widespread environmental pollutant, and their ecological and environmental health risks have gradually attracted attention. To reveal the toxicity characteristics of these compounds, ten PAEs were selected as research objects to establish a cell model. CCK-8 was used to determine cell viability, Western blots were used to determine the content of Nrf₂ in HepG2, and the LD₅₀ collected for the 13 PAEs administered to rats. On this basis, 3D-QSAR models of IC₅₀, LD₅₀ and Nrf2 were established. The experimental results showed that as the time of PAEs exposure increased (24, 48 and 72 h), cell viability gradually decreased. The test concentration (62.5 /125/250 μM) of PAEs exposed for 48 h could significantly increase the content of Nrf2, and the 1000 μM PAEs could inhibit the content of Nrf2. The model is relatively stable and predicts well that the introduction of large and hydrophobic groups may significantly affect the toxic effects of PAEs on cells. The present study provided a potential tool for predicting the LD50 and Nrf2 of new PAEs, and provide a reference for the design of new less toxic PAEs in the future.

Computer modeling of some anti-breast cancer compounds

The research was aimed at exploring the biological activities of novel series of β-lactam derivatives against MCF-7 breast cancer cell lines via computer modeling such as quantitative structure-activity relationship (QSAR), designing new compounds and analyzing the drug likeliness of designed compounds. The QSAR model was highly robust as it also conforms to the least minimum requirement for QSAR model from the statistical assessments with a correlation coefficient squared (R2) of 0.8706, correlation coefficient adjusted squared (R2adj) of 0.8411, and cross-validation coefficient (Q2) of 0.7844. The external validation of R2pred was calculated as 0.6083 for model 4. The model parameters (MATS5i and MATS1s) were used in designing new derivative compounds with higher potency against estrogen-positive breast cancer. The pharmacokinetics test on the restructured compounds revealed that all the compounds passed the drug likeness test and they could further proceed to clinical trials. These reveal a breakthrough in medicine, in the research for breast cancer drug with higher effectiveness against the MCF-7 cell line.

Do we need different machine learning algorithms for QSAR modeling? A comprehensive assessment of 16 machine learning algorithms on 14 QSAR data sets

Although a wide variety of machine learning (ML) algorithms have been utilized to learn quantitative structure-activity relationships (QSARs), there is no agreed single best algorithm for QSAR learning. Therefore, a comprehensive understanding of the performance characteristics of popular ML algorithms used in QSAR learning is highly desirable. In this study, five linear algorithms [linear function Gaussian process regression (linear-GPR), linear function support vector machine (linear-SVM), partial least squares regression (PLSR), multiple linear regression (MLR) and principal component regression (PCR)], three analogizers [radial basis function support vector machine (rbf-SVM), K-nearest neighbor (KNN) and radial basis function Gaussian process regression (rbf-GPR)], six symbolists [extreme gradient boosting (XGBoost), Cubist, random forest (RF), multiple adaptive regression splines (MARS), gradient boosting machine (GBM), and classification and regression tree (CART)] and two connectionists [principal component analysis artificial neural network (pca-ANN) and deep neural network (DNN)] were employed to learn the regression-based QSAR models for 14 public data sets comprising nine physicochemical properties and five toxicity endpoints. The results show that rbf-SVM, rbf-GPR, XGBoost and DNN generally illustrate better performances than the other algorithms. The overall performances of different algorithms can be ranked from the best to the worst as follows: rbf-SVM > XGBoost > rbf-GPR > Cubist > GBM > DNN > RF > pca-ANN > MARS > linear-GPR ≈ KNN > linear-SVM ≈ PLSR > CART ≈ PCR ≈ MLR. In terms of prediction accuracy and computational efficiency, SVM and XGBoost are recommended to the regression learning for small data sets, and XGBoost is an excellent choice for large data sets. We then investigated the performances of the ensemble models by integrating the predictions of multiple ML algorithms. The results illustrate that the ensembles of two or three algorithms in different categories can indeed improve the predictions of the best individual ML algorithms.

Cloud 3D-QSAR: a web tool for the development of quantitative structure-activity relationship models in drug discovery

Effective drug discovery contributes to the treatment of numerous diseases but is limited by high costs and long cycles. The Quantitative Structure-Activity Relationship (QSAR) method was introduced to evaluate the activity of a large number of compounds virtually, reducing the time and labor costs required for chemical synthesis and experimental determination. Hence, this method increases the efficiency of drug discovery. To meet the needs of researchers to utilize this technology, numerous QSAR-related web servers, such as Web-4D-QSAR and DPubChem, have been developed in recent years. However, none of the servers mentioned above can perform a complete QSAR modeling and supply activity prediction functions. We introduce Cloud 3D-QSAR by integrating the functions of molecular structure generation, alignment, molecular interaction field (MIF) computing and results analysis to provide a one-stop solution. We rigidly validated this server, and the activity prediction correlation was R2 = 0.934 in 834 test molecules. The sensitivity, specificity and accuracy were 86.9%, 94.5% and 91.5%, respectively, with AUC = 0.981, AUCPR = 0.971. The Cloud 3D-QSAR server may facilitate the development of good QSAR models in drug discovery. Our server is free and now available at http://chemyang.ccnu.edu.cn/ccb/server/cloud3dQSAR/ and http://agroda.gzu.edu.cn:9999/ccb/server/cloud3dQSAR/.

Computer-aided Drug Design Investigations for Benzothiazinone Derivatives Against Tuberculosis

BACKGROUND

Tuberculosis (Mycobacterium tuberculosis) is an infectious bacterial disease with the highest levels of mortality worldwide, presenting numerous cases of resistance. In silico studies, which elaborate chemical and biological models in computational tools and make it possible to interpret molecular characteristics, are among the methods used in the search for new drugs.

OBJECTIVE

In this perspective, our aim was to use QSAR and molecular modeling to propose possible pharmacophores from benzothiazinone derivatives.

METHODS

In this study, a set of 69 benzothiazinone derivatives, together with computational tools such as molecular descriptor analysis in chemometrics, metabolic prediction, and molecular coupling to 4 proteins: DprE1, InhA, PS, and DHFR important for the bacillus were investigated.

RESULTS

The chemometric model computed in the Volsurf+ program presented good predictive values for both amphiphilicity and molecular volume. These are essential for biological activity. Metabolites from the cytochrome isoforms CYP3A4 and 2D6 interactions revealed coupling divergences which, noting that the metabolites did not present changes to the QSAR proposed pharmacophore structures, may be due to the reaction medium and existing differences in the benzothiazinone structures. Similarly, molecular docking with the four TB enzymes presented good interactions for the more active compounds. The fragments found using QSAR (being essential for biological activity) also presented as being essential for ligand-protein site interactions.

CONCLUSION

From the benzothiazinone derivative series evaluated, compound 11026134 presented the best profile in all study analyses, noting that the trifluoromethyl, nitro group, and piperazine fragment with aliphatic hydrocarbon groups are likely pharmacophores for the benzothiazinones studied.

Ecotoxicological QSAR modelling of organic chemicals against Pseudokirchneriella subcapitata using consensus predictions approach

The present study provides robust consensus quantitative structure-activity relationship (QSAR) models developed from 334 organic chemicals covering a wide chemical domain for the prediction of effective concentrations of chemicals for 50% and 10% inhibition of algal growth. Only 2D descriptors with definite physicochemical meaning were employed for QSAR model building, whereas development, validation and interpretation were achieved following the strict Organization for Economic Co-operation and Development (OECD) recommended guidelines. Genetic algorithm along with stepwise approach was used in feature selection while the final QSAR models were derived using partial least squares regression technique. The applicability domain of the developed models was also checked. The obtained consensus models were then used to predict 64 organic chemicals having no definite observed responses while the confidence of predictions was checked by the 'prediction reliability indicator' tool. The developed models should be applicable for data gap filling in case of new or untested organic chemicals provided they fall within the domain of the model and can also be implemented to design safer alternatives to the environment.