Is regression through origin useful in external validation of QSAR models?

In silico insights into design of novel VEGFR-2 inhibitors: SMILES-based QSAR modelling, and docking studies on substituted benzo-fused heteronuclear derivatives

Eight QSAR models (M1-M8) were developed from a dataset of 118 benzo-fused heteronuclear derivatives targeting VEGFR-2 by Monte Carlo optimization method of CORALSEA 2023 software. Models were generated with hybrid optimal descriptors using both SMILES and Graphs with zero- and first-order Morgan extended connectivity index from a training set of 103 derivatives. All statistical parameters for model validation were within the prescribed limits, establishing the models to be robust and of excellent quality. Among all models, split-2 of M5 was the best-fit as reflected by   r v a lidation 2 ,   Q v a lidation 2 and MAE . Mechanistic interpretation of this model assisted the identification of structural descriptors as promoters and hinderers for VEGFR-2 inhibition. These descriptors were utilized to design novel VEGFR-2 inhibitors (YS01-YS07) by bringing modifications in compound MS90 in the dataset. Docking of all designed compounds, MS90 and sorafenib with VEGFR-2 binding site revealed favourable binding interactions. Docking score of YS07 was higher than that of MS90 and sorafenib. Molecular dynamics simulation study revealed sustained interactions of YS07 with key amino acids of VEGFR-2 at a run time of 100 ns. This study concludes the development of a best fit QSAR model which can assist the design of new anticancer agents targeting VEGFR-2.

QSAR models for the ozonation of diverse volatile organic compounds at different temperatures

In order to assess the fate and persistence of volatile organic compounds (VOCs) in the atmosphere, it is necessary to determine their oxidation rate constants for their reaction with ozone (kO3). However, given that experimental values of kO3 are only available for a few hundred compounds and their determination is expensive and time-consuming, developing predictive models for kO3 is of great importance. Thus, this study aimed to develop reliable quantitative structure-activity relationship (QSAR) models for 302 values of 149 VOCs across a broad temperature range (178-409 K). The model was constructed based on the combination of a simplified molecular-input line-entry system (SMILES) and temperature as an experimental condition, namely quasi-SMILES. In this study, temperature was incorporated in the models as an independent feature. The hybrid optimal descriptor generated from the combination of quasi-SMILES and HFG (hydrogen-filled graph) was used to develop reliable, accurate, and predictive QSAR models employing the CORAL software. The balance between the correlation method and four different target functions (target function without considering IIC or CII, target function using each IIC or CII, and target function based on the combination of IIC and CII) was used to improve the predictability of the QSAR models. The performance of the developed models based on different target functions was compared. The correlation intensity index (CII) significantly enhanced the predictability of the model. The best model was selected based on the numerical value of Rm2 of the calibration set (split #1, Rtrain2 = 0.9834, Rcalibration2 = 0.9276, Rvalidation2 = 0.9136, and calibration = 0.8770). The promoters of increase/decrease for log kO3 were also computed based on the best model. The presence of a double bond (BOND10000000 and $10 000 000 000), absence of halogen (HALO00000000), and the nearest neighbor codes for carbon equal to 321 (NNC-C⋯321) are some significant promoters of endpoint increase.

Multivariate QSAR, similarity search and ADMET studies based in a set of methylamine derivatives described as dopamine transporter inhibitors

The dopamine transporter (DAT), responsible for the regulation of dopaminergic neurotransmission, is implicated in the etiology of several neuropsychiatric disorders which, in turn, have contributed to high rates of disability and numerous deaths in recent years, significantly impacting the global health system. Although the research for new drugs for the treatment of neuropsychiatric disorders has evolved in recent years, the availability of DAT-selective drugs that do not generate the same psychostimulant effects observed in drugs of abuse remains scarce. Therefore, we performed a QSAR study based on a dataset of 36 methylamine derivatives described as DAT inhibitors. The model was obtained based only in descriptors derived from 2D structures, and it was validated and generated satisfactory results considering the metrics used for internal and external validation. Subsequently, a virtual screening step also based on 2D similarity was performed, where it was possible to identify a total of 1157 compounds. After a series of reductions of the set using toxicity filters, applicability domain evaluation, and pharmacokinetic properties in silico assessment, seven hit compounds were selected as the most promising to be used, in future studies, as new scaffolds for the development of new DAT inhibitors.

Machine learning modeling to identify affinity improved biobetter anticancer drug trastuzumab and the insight of molecular recognition of trastuzumab towards its antigen HER2

In the present study, a machine learning (ML) model was developed to predict the epistatic phenomena of combination mutants to improve the anticancer antibody-drug trastuzumab's binding affinity towards its antigen human epidermal growth factor receptor 2 (HER2). An ML algorithm, Support Vector Regression (SVR) was used to develop ML models with a data set consists of 193 affinity values of single mutants of trastuzumab and its associated various amino acid sequence derived descriptors. The subset selection of descriptors and SVR hyperparameters were done using the Genetic Algorithm (GA) within the SVR and the wrapper approach called GA-SVR. A 100 evolutionary cycles of GA produced the best 100 probable GA-SVR models based on their fitness score (Q2) estimated using a stratified 5 fold cross-validation procedure. The final ML model found to be highly predictive of test data set of six combination mutants and one single mutant with Rpre2 = 0.71. The analysis of descriptors in the ML model highlighted the importance of mutant induced secondary structural variation causes the binding affinity variation of the trastuzumab. The same was verified using a short 20 ns and a long 100 ns in duplicate molecular dynamics simulation of a wild and mutant variant of trastuzumab. The secondary structure induced affinity change due to mutations in the CDR-H3 is a novel insight that came out of this study. That should help rational mutant selection to develop a biobetter trastuzumab with a multifold improved binding affinity into the market quickly.Communicated by Ramaswamy H. Sarma.

Use of quasi-SMILES to build models based on quantitative results from experiments with nanomaterials

Quasi-SMILES deviate from traditional SMILES (simplified molecular input-line entry system) by the extension of additional symbols that encode for conditions of an experiment. Descriptors calculated with SMILES are useful for the development of quantitative structure-property/activity relationships (QSPRs/QSARs), while descriptors calculated with quasi-SMILES can be useful for the development of quantitative models of experimental results obtained under different conditions. Here, this approach has been applied for the development of generalized models using aquatic nanotoxicity data (i.e., related to fish and daphnia). The statistical quality of the above models (pLC50) is quite good with a determination coefficient for the external validation set ranging from 0.62 to 0.71 and RMSE ranging from 0.58 to 0.60. The principle of the approach includes splitting the experimental data into three random distributions defining training, calibration, and validation sets.

Comparison of various methods for validity evaluation of QSAR models

BACKGROUND

Quantitative structure-activity relationship (QSAR) modeling is one of the most important computational tools employed in drug discovery and development. The external validation of QSAR models is the main point to check the reliability of developed models for the prediction activity of not yet synthesized compounds. It was performed by different criteria in the literature.

METHODS

In this study, 44 reported QSAR models for biologically active compounds reported in scientific papers were collected. Various statistical parameters of external validation of a QSAR model were calculated, and the results were discussed.

RESULTS

The findings revealed that employing the coefficient of determination (r²) alone could not indicate the validity of a QSAR model. The established criteria for external validation have some advantages and disadvantages which should be considered in QSAR studies.

CONCLUSION

This study showed that these methods alone are not only enough to indicate the validity/invalidity of a QSAR model.

CORAL: Development of a hybrid descriptor based QSTR model to predict the toxicity of dioxins and dioxin-like compounds with correlation intensity index and consensus modelling

In the present study, ninety-five halogenated dioxins and related chemicals (dibenzo-p-dioxins, dibenzofurans, biphenyls, and naphthalene) with endpoint pEC₅₀ were used to develop twelve quantitative structure toxicity relationship (QSTR) models using inbuilt Monte Carlo algorithm of CORAL software. The hybrid optimal descriptor of correlation weights (DCW) using a combination of SMILES and HSG (hydrogen suppressed graph) was employed to generate QSTR models. Three target functions i.e. TF₁ (W_IIC=W_CII=0), TF₂ (W_IIC= 0.3 & W_CII=0) and TF₃ (W_IIC= 0.0 &W_CII=0.3) were employed to develop robust QSTR models and the statistical outcomes of each target function were compared with each other. The correlation intensity index (CII) was found a reliable benchmark of the predictive potential for QSTR models. The numerical value of the determination coefficient of the validation set of split 1 computed by TF₃ was found highest (R_Valid²=0.8438). The fragments responsible for the toxicity of dioxins and related chemicals were also identified in terms of the promoter of increase/decrease for pEC₅₀. Three random splits (Split 1, Split 2 and Split 4) were selected for the extraction of the promoter of increase/decrease for pEC₅₀. In the last, consensus modelling was performed using the intelligent consensus tool of DTC lab (https://dtclab.webs.com/software-tools). The original consensus model, which was created by combining four distinct models employing the split 4 arrangement, was more predictive for the validation set and the numerical value of the determination coefficient of the test set (validation set) was increased from 0.8133 to 0.9725. For the validation set of split 4, the mean absolute error (MAE 100%) was also lowered from 0.513 to 0.2739.

Quantitative structure-toxicity relationship models for predication of toxicity of ionic liquids towards Leukemia rat cell line IPC-81 based on index of ideality of correlation

The application of ion liquids (ILs) as green solvents has attracted the attention of the scientific community. However, ILs may play the role of toxins. Even though ionic liquids may assist to minimise air pollution, but their discharge into aquatic ecosystems might result in significant water pollution due to their potential toxicity and inaccessibility to biodegradation. Recently, more attention has been paid to the toxicity of ILs on plants, bacteria, and humans. Here, a quantitative structure-toxicity relationship study (QSTR) based on the Monte Carlo method of CORAL software has been applied to estimate the logarithm of the half-maximal effective concentration of toxicity of ILs against leukemia rat cell line IPC-81 (logEC₅₀). A hybrid optimal descriptor is used to build QSTR models for a large set of 304 diverse ILs including ammonium, imidazolium, morpholinium, phosphonium, piperidinium, pyridinium, pyrrolidinium, quinolinium, sulfonium, and protic ILs. The SMILES notations of Ils are utilized to compute the descriptor correlation weight (DCW). Four splits are made from the whole dataset and each split is randomly divided into four sets (training subsets and validation set). The index of ideality of correlation (IIC) is applied to evaluate the authenticity and robustness of the QSTR models. A QSTR model with statistical parameters R²=0.85, CCC =0.92, Q²=0.84, and MAE =0.25 for the validation set of the best split is considered as a prime model. The outliers and promoters of increase/decrease of logEC₅₀ are extracted and the mechanistic interpretation of effective descriptors for the model is also offered.HighlightsGlobal SMILES-based QSAR model was developed to predict the toxicity of ILs.The CORAL software is used to model the ILs toxicity on IPC-81 leukemia rat cell lineIIC is tested as a criterion of predictive potential.The toxicological effects of ILs are discussed based on the proposed model.

The Monte Carlo approach to model and predict the melting point of imidazolium ionic liquids using hybrid optimal descriptors

Ionic liquids (ILs) have captured intensive attention owing to their unique properties such as high thermal stability, negligible vapour pressure, high dissolution capacity and high ionic conductivity as well as their wide applications in various scientific fields including organic synthesis, catalysis, and industrial extraction processes. Many applications of ionic liquids (ILs) rely on the melting point (T_m). Therefore, in the present manuscript, the melting points of imidazolium ILs are studied employing a quantitative structure–property relationship (QSPR) approach to develop a model for predicting the melting points of a data set of imidazolium ILs. The Monte Carlo algorithm of CORAL software is applied to build up a robust QSPR model to calculate the values T_m of 353 imidazolium ILs. Using a combination of SMILES and hydrogen-suppressed molecular graphs (HSGs), the hybrid optimal descriptor is computed and used to generate the QSPR models. Internal and external validation parameters are also employed to evaluate the predictability and reliability of the QSPR model. Four splits are prepared from the dataset and each split is randomly distributed into four sets i.e. training set (≈33%), invisible training set (≈31%), calibration set (≈16%) and validation set (≈20%). In QSPR modelling, the numerical values of various statistical features of the validation sets such as R_Validation², Q_Validation², and IIC_Validation are found to be in the range of 0.7846–0.8535, 0.7687–0.8423 and 0.7424–0.8982, respectively. For mechanistic interpretation, the structural attributes which are responsible for the increase/decrease of T_m are also extracted.

The melting points of imidazolium ILs are studied employing a quantitative structure–property relationship (QSPR) approach to develop a model for predicting the melting points of a data set of imidazolium ILs.

The unreliability of the reliability criteria in the estimation of QSAR for skin sensitivity: A pun or a reliable law?

Some new products, which include common personal-care products, drugs, household items, can be hazardous in aspect personal care products/cosmetics and their ingredients (i.e. the above can effect human skin). International organizations (e.g. the Organisation for Economic Co-operation and Development-OECD) recommend evaluating individual ingredients when assessing the safety of personal care or cosmetic products. Thus, checking up that "popular at the market" substances are non-toxic, do not penetrate into or through normal or compromised human skin, and therefore, pose no risk to human health is an essential element of modern toxicology. The development of reliable models of toxicological endpoints is a tool to carry out the above checking up via quantitative structure-activity relationships (QSARs). The reliability of the QSAR is the current task of mathematical statistics. Recently, the index of ideality of correlation (IIC) and correlation intensity index (CII) were suggested as criteria of predictive potential (i.e. reliability) of QSAR-models. Here, the abilities of these criteria were studied for the case of building up models for skin sensitivity (LLNA, local lymph node assay). Computational experiments have confirmed that the IIC demonstrates an obvious ability to improve the predictive potential of models of skin sensitization. The applying of the CII for the case of skin sensitization also improves the quality of the model. However, the best models for skin sensitization were observed if the above-mentioned criteria are applied jointly (n = 268; R² = 0.60; RMSE = 0.63).

Correlation intensity index: Building up models for mutagenicity of silver nanoparticles

Nanomaterials become significant component of economics. Consequently, nanomaterials become object of environmental sciences. There is a traditional list of endpoints which are indicators of the ecological risk. Mutagenicity is one of important component in this list. The quasi-SMILES approach, that in contrast to majority of work dedicated to modelling behaviour of nanomaterials gives possibility to consider experimental conditions as well as other circumstances which can impact the behaviour of nanomaterials is suggested. This is carried out via so-called quasi-SMILES. The quasi-SMILES is a line on of codes that contains all the above available eclectic data. Modelling process aimed to build up a model involves Correlation Intensity Index (CII) that is a new criterion of predictive potential of models. The scheme of calculation of CII is described in this work in the first time. The applying of CII together with Index of Ideality Correlation (IIC) in modelling of mutagenicity of silver nanoparticles by the Monte Carlo method using the CORAL software (http://www.insilico.eu/coral) indicates that application of the CII improves the predictive potential of these models for three random splits into the training set (75%) and validation set (25%).

Quantitative structure-property relationship of the photoelectrochemical oxidation of phenolic pollutants at modified nanoporous titanium oxide using supervised machine learning

Here we report on an advanced photoelectrochemical (PEC) oxidation of 22 phenolic pollutants based on modified nanoporous TiO2, which was directly grown on a titanium substrate electrochemically. Their degradation rate constants were experimentally determined and their physicochemical properties were computaionally calculated. The quantitative structure-property relationship (QSPR) was elucidated by employing multiple linear regression (MLR) method. A supervised machine learning approach was employed to build QSPR models. The high predictive abilities of the QSPR model were validated via leave-one-out (LOO) method and a strict regimen of statistical validation tests. The significant descriptors identified in the QSPR Model for the phenolic compounds were also assessed using a typical dye pollutant Rhodamine B, further confirming the high effectiveness and predictability of the optimized model. Our study has shown that the integrated effect of the structural, hydrophobic and topological properties along with electronic property should be considered in order to design an efficient PEC catalytic approach for environmental applications.

Molecular modeling studies of benzothiophene-containing derivatives as promising selective estrogen receptor downregulators: a combination of 3D-QSAR, molecular docking and molecular dynamics simulations

Selective estrogen receptor downregulators (SERDs) for the treatment of positive breast cancer can act both as estrogen alpha receptor (ERα) antagonists and degraders. In this study, the optimal antagonist models (CoMFA-A, q² = 0.660, r² = 0.996; CoMSIA-A, q² = 0.728, r² = 0.992) and degrader models (CoMFA-D, q² = 0.850, r² = 0.996; CoMSIA-D, q² = 0.719, r² = 0.995) of a series of potent benzothiophene-containing SERDs were constructed to explore the three-dimensional quantitative structure-activity relationship. Internal and external validation indicated that all models exhibited good applicability, high predictive ability and robustness. Contour maps revealed the relationships between the essential structural features and antagonistic and degradation activities. Additionally, molecular docking, molecular dynamics and free energy calculation studies were further performed to investigate the detailed binding mode. Results indicated that several key residues, ARG394, GLU353, PHE404 and ILE424, were crucial for the stability of the ligand binding domain. The hydrophobic, electrostatic and Van der Waals interactions played significant effect on the binding affinity. Finally, ten novel compounds were designed based on above findings, where the predicted activity of compound D8 was equivalent to that of the compound LSZ102. 3D-QSAR, ADMET and bioavailability predictions indicated that all designed compounds with good predicted activity, good physicochemical and bioavailability could be potential candidates for SERDs. These results and combinations of computational methods provided guidance for the rational drug design of novel potential SERDs.Communicated by Ramaswamy H. Sarma.

Whether the Validation of the Predictive Potential of Toxicity Models is a Solved Task?

Different kinds of biological activities are defined by complex biochemical interactions, which are termed as a "mathematical function" not only of the molecular structure but also for some additional circumstances, such as physicochemical conditions, interactions via energy and information effects between a substance and organisms, organs, cells. These circumstances lead to the great complexity of prediction for biochemical endpoints, since all "details" of corresponding phenomena are practically unavailable for the accurate registration and analysis. Researchers have not a possibility to carry out and analyse all possible ways of the biochemical interactions, which define toxicological or therapeutically attractive effects via direct experiment. Consequently, a compromise, i.e. the development of predictive models of the above phenomena, becomes necessary. However, the estimation of the predictive potential of these models remains a task that is solved only partially. This mini-review presents a collection of attempts to be used for the above-mentioned task, two special statistical indices are proposed, which may be a measure of the predictive potential of models. These indices are (i) Index of Ideality of Correlation; and (ii) Correlation Contradiction Index.

Aug-MIA-QSAR based strategy in bioactivity prediction of a series of flavonoid derivatives as HIV-1 inhibitors

Multivariate image analysis-quantitative structure-activity relationship (MIA-QSAR) is a simple and quite accessible QSAR method for predicting biological activities of compounds based on two-dimensional image analysis. Aug-MIA-QSAR is a modified version of multivariate image analysis, where the atoms in 2D chemical structures were augmented (labelled by assigning specific colours). This study focuses on efficiently constructing such prediction models using a dataset of flavonoid derivatives possessing human immunodeficiency virus - 1 inhibition. The models were constructed by partial least square regression using non-linear iterative partial least square (NIPALS) algorithm and linearized by identifying an optimum number of seven latent variables. A leave-one-out cross validation (LOOCV) helped to verify the actual and predicted data. The two multivariate methods were compared and analysed to identify the most suitable method.

MIA-QSAR based model for bioactivity prediction of flavonoid derivatives as acetylcholinesterase inhibitors

Alzheimer's disease is a common form of dementia, which considered to be a major health concern. Multivariate Image Analysis - Quantitative Structure-Activity Relationship (MIA-QSAR) is a simple and quite accessible QSAR method for predicting biological activities of unstudied compounds based on 2D image analysis. This study focuses on constructing an efficient QSAR model using a dataset of 52 flavonoid derivatives (substituted with amino-alkyl, alkoxy, alkyl-amines, and piperidine groups) as active compounds against acetylcholinesterase inhibitors (AChE). The model was constructed by PLS (Partial Least Square) using NIPALS (Non-Linear iterative Partial Least Square) algorithm. The comparable values obtained from calibration of training set using five latent variables (R² = 0.955) and external validation of test set (Q² = 0.948) confirmed the precision in the prediction of bioactivities for the set of flavonoid derivatives used in designing the model.

Chemometric-assisted method development in hydrophilic interaction liquid chromatography: A review

With an enormous growth in the application of hydrophilic interaction liquid chromatography (HILIC), there has also been significant progress in HILIC method development. HILIC is a chromatographic method that utilises hydro-organic mobile phases with a high organic content, and a hydrophilic stationary phase. It has been applied predominantly in the determination of small polar compounds. Theoretical studies in computer-aided modelling tools, most importantly the predictive, quantitative structure retention relationship (QSRR) modelling methods, have attracted the attention of researchers and these approaches greatly assist the method development process. This review focuses on the application of computer-aided modelling tools in understanding the retention mechanism, the classification of HILIC stationary phases, prediction of retention times in HILIC systems, optimisation of chromatographic conditions, and description of the interaction effects of the chromatographic factors in HILIC separations. Additionally, what has been achieved in the potential application of QSRR methodology in combination with experimental design philosophy in the optimisation of chromatographic separation conditions in the HILIC method development process is communicated. Developing robust predictive QSRR models will undoubtedly facilitate more application of this chromatographic mode in a broader variety of research areas, significantly minimising cost and time of the experimental work.

Beware of R(2): Simple, Unambiguous Assessment of the Prediction Accuracy of QSAR and QSPR Models

The statistical metrics used to characterize the external predictivity of a model, i.e., how well it predicts the properties of an independent test set, have proliferated over the past decade. This paper clarifies some apparent confusion over the use of the coefficient of determination, R(2), as a measure of model fit and predictive power in QSAR and QSPR modeling. R(2) (or r(2)) has been used in various contexts in the literature in conjunction with training and test data for both ordinary linear regression and regression through the origin as well as with linear and nonlinear regression models. We analyze the widely adopted model fit criteria suggested by Golbraikh and Tropsha ( J. Mol. Graphics Modell. 2002 , 20 , 269 - 276 ) in a strict statistical manner. Shortcomings in these criteria are identified, and a clearer and simpler alternative method to characterize model predictivity is provided. The intent is not to repeat the well-documented arguments for model validation using test data but rather to guide the application of R(2) as a model fit statistic. Examples are used to illustrate both correct and incorrect uses of R(2). Reporting the root-mean-square error or equivalent measures of dispersion, which are typically of more practical importance than R(2), is also encouraged, and important challenges in addressing the needs of different categories of users such as computational chemists, experimental scientists, and regulatory decision support specialists are outlined.

The rm2 metrics and regression through origin approach: reliable and useful validation tools for predictive QSAR models (Commentary on 'Is regression through origin useful in external validation of QSAR models?')

Quantitative structure-activity relationship (QSAR) is an in silico technique which can be used in drug discovery, environmental fate modeling, property and toxicity prediction of chemical entities and regulatory toxicology. The predictive potential of a QSAR model is judged from various validation metrics in order to evaluate how well it is capable to predict endpoint values of new untested compounds. The rm2 group of metrics is one of the stringent validation metrics currently used by the QSAR fraternity in different reports. We scrutinized a recently published paper which raised an issue that the constructed criteria based on regression through origin (RTO) are not optimal and there is a significant difference in the rm2 metrics values computed from different statistical software packages. According to our point of view, the conclusion drawn in this paper appears to be misleading. Any inconsistency in the software algorithms has nothing to do with the calculation of rm2 metrics, as such computation is not limited by the use of any specific software, rather it depends only on fundamental mathematical formulae that are well established. However, it is a concern to the QSAR users that Excel and SPSS can return different results for the metrics using the RTO method. Thus, a proper validation of the software tool is required before its use for computation of any validation metric.