An Update of the BCF QSAR Model Based on Theoretical Molecular Descriptors

Classification-based QSARs for predicting dietary biomagnification in fish

The assessment of bioaccumulation is an important step to describe the environmental behaviour and the potential risk due to exposure to potentially hazardous chemicals. In the last two decades, several in silico tools have been made available to predict bioconcentration, which is commonly used to assess bioaccumulation in risk assessment frameworks all over the world. However, only a few QSAR studies address the prediction of the biomagnification factor (BMF), which describes the accumulation of chemicals into organisms due to exposure through the diet. No classification models are currently available to this end. In this work, we developed classification QSARs to predict classes based on dietary biomagnification, using three different classifiers (i.e. LDA, ANN and RF). We started from a recently published dataset that includes more than 300 curated dietary BMF values measured in fish. The new models have high-quality performances (accuracy in fitting: from 94 to 96%; accuracy in prediction from 84 to 86%). The good performances of the here proposed QSARs confirm the quality of the original input data and highlight the importance of data curation and data sharing to support the development of new in silico approaches to assist risk assessment and chemicals screening.

QSARINS-Chem standalone version: A new platform-independent software to profile chemicals for physico-chemical properties, fate, and toxicity

The new software QSARINS-Chem standalone version is a multiplatform tool, freely downloadable, for the in silico profiling of multiple properties and activities of organic chemicals. This software, which is based on the concept of the QSARINS-chem module embedded in the QSARINS software, has been fully redesigned and redeveloped in the Java™ language. In addition to a selection of models included in the old module, the new software predicts biotransformation rates and aquatic toxicities of pharmaceuticals and personal care products in multiple organisms, and offers a suite of tools for the analysis of predictions. Furthermore, a comprehensive and transparent database of molecular structures is provided. The new QSARINS-Chem standalone version is an informative and solid tool, which is useful to support the assessment of the potential hazard and risks related to organic chemicals and is dedicated to users which are interested in the application of QSARs to generate reliable predictions.

In silico local QSAR modeling of bioconcentration factor of organophosphate pesticides

The persistent and accumulative nature of the pesticide of indiscriminate use emerged as ecotoxicological hazards. The bioconcentration factor (BCF) is one of the key elements for environmental assessments of the aquatic compartment. Limitations of prediction accuracy of global model facilitate the use of local predictive models in toxicity modeling of emerging compounds. The BCF data of diverse organophosphate (n = 55) was collected from the Pesticide Properties Database and used as a model data set in the present study to explore physicochemical properties and structural alert concerning BCF. The structures were downloaded from Pubchem, ChemSpider database. Two splitting techniques (biological sorting and structure-based) were used to divide the whole dataset into training and test set compounds. The QSAR study was carried out with two-dimensional descriptors (2D) calculated from PaDEL by applying genetic algorithm (GA) as chemometric tools using QSARINS software. The models were statistically robust enough both internally as well as externally (Q2: 0.709-0.722, Q2 Ext: 0.717-0.903, CCC: 0.857-0.880). Overall molecular mass, presence of fused, and heterocyclic ring with electron-withdrawing groups affect the BCF value. The developed models reflected extended applicability domain (AD) and reliable predictions than the reported models for the studied chemical class. Finally, predictions of unknown organophosphate pesticides and the toxic nature of unknown organophosphate pesticides were commented on. These findings may be useful for the scientific community in prioritizing high potential pesticides of organophosphate class.

Amino acid derivative reactivity assay-organic solvent reaction system: A novel alternative test for skin sensitization capable of assessing highly hydrophobic substances

The amino acid derivative reactivity assay (ADRA) is an in chemico alternative to animal testing that focuses on protein binding. The ADRA is a skin sensitization test that solves problems associated with the direct peptide reactivity assay. However, when utilizing the ADRA to evaluate highly hydrophobic substances with octanol/water partition coefficients (logKow) of >6, the test substances may not dissolve in the reaction solution, which can prevent the accurate assessment of skin sensitization. Therefore, we developed the ADRA-organic solvent (ADRA-OS) reaction system, which is a novel skin sensitization test that enables the assessment of highly hydrophobic substances with a logKow of >6. We discovered that the organic solvent ratio, the triethylamine concentration, and the ethylenediaminetetraacetic acid disodium salt dihydrate concentration participate in reactions with the nucleophile N-(2-(1-naphthyl)acetyl)-l-cysteine (NAC) and sensitizers that are used in ADRA and in stabilizing NAC. Thus, we determined the optimal reaction composition of the ADRA-OS according to L₉ (3³ ) orthogonal array experiments. Using this test, we assessed 14 types of highly hydrophobic substances. When we compared the results with ADRA, we found that ADRA-OS reaction system has high solubility for highly hydrophobic substances and that it has a high predictive capacity (sensitivity: 63%, specificity: 100%, accuracy: 79%). The implication of the results is that the novel ADRA-OS reaction system should provide a useful method for assessing the skin sensitization of highly hydrophobic substances with a logKow of >6.

Principles of QSAR Modeling

At the end of her academic career, the author summarizes the main aspects of QSAR modeling, giving comments and suggestions according to her 23 years' experience in QSAR research on environmental topics. The focus is mainly on Multiple Linear Regression, particularly Ordinary Least Squares, using a Genetic Algorithm for variable selection from various theoretical molecular descriptors, but the comments can be useful also for other QSAR methods. The need for rigorous validation, also external, and for applicability domain check to guarantee predictivity and reliability of QSAR models is particularly highlighted. The commented approach is the “predictive” one, based on chemometrics, and is usefully applied to the prioritization of environmental pollutants. All the discussed points and the author's ideas are implemented in the software QSARINS, as a legacy to the QSAR community.

QSAR modelling study of the bioconcentration factor and toxicity of organic compounds to aquatic organisms using machine learning and ensemble methods

Bioconcentration factors and median lethal concentrations (LC₅₀s) are important when assessing risks posed by organic pollutants to aquatic ecosystems. Various quantitative structure-activity relationship models have been developed to predict bioconcentration factors and classify acute toxicity. In the study, we developed a regression model using Recursive Feature Elimination (RFE) method combined with the Support Vector Machine (SVM) algorithm. We calculated 2D molecular descriptors from a dataset containing 450 diverse chemicals in our regression model. Then we built three ensemble models using three machine learning algorithms and calculated 12 molecular fingerprints from a dataset containing 400 diverse chemicals in our classification models. In the regression model, the R² and R_pred² for the regression model were 0.860 and 0.757, respectively. Other parameters indicated that the regression model made good predictions and could efficiently predict a new set of compounds following standards set by Golbraikh, Tropsha, and Roy. In the classification models, the ensemble-SVM classification model gave an overall accuracy, sensitivity, specificity, and area under the receiver operating characteristic curve of 92.2, 95.1, 86.0, and 0.965, respectively, in a five-fold cross-validation and of 87.3, 92.6, 76.0, and 0.940, respectively, in an external validation. These parameters indicated that our ensemble-SVM model was more stable and gave more accurate predictions than previous models. The model could therefore be used to effectively predict aquatic toxicity and assess risks posed to aquatic ecosystems. We identified several structures most relevant to acute aquatic toxicity through predictions made by the two types of models, and this information may be important to aquatic toxicology experiments and aquatic system management.

3D matters! 3D-RISM and 3D convolutional neural network for accurate bioaccumulation prediction

In this work, we present a new method for predicting complex physical-chemical properties of organic molecules. The approach utilizes 3D convolutional neural network (ActivNet4) that uses solvent spatial distributions around solutes as input. These spatial distributions are obtained by a molecular theory called three-dimensional reference interaction site model. We have shown that the method allows one to achieve a good accuracy of prediction of bioconcentration factor which is difficult to predict by direct application of methods of molecular theory or simulations. Our research demonstrates that combination of molecular theories with modern machine learning approaches can be effectively used for predicting properties that are otherwise inaccessible to purely theory-based models.

QSAR modeling of cumulative environmental end-points for the prioritization of hazardous chemicals

The hazard of chemicals in the environment is inherently related to the molecular structure and derives simultaneously from various chemical properties/activities/reactivities. Models based on Quantitative Structure Activity Relationships (QSARs) are useful to screen, rank and prioritize chemicals that may have an adverse impact on humans and the environment. This paper reviews a selection of QSAR models (based on theoretical molecular descriptors) developed for cumulative multivariate endpoints, which were derived by mathematical combination of multiple effects and properties. The cumulative end-points provide an integrated holistic point of view to address environmentally relevant properties of chemicals.

Modeling adsorption of brominated, chlorinated and mixed bromo/chloro-dibenzo-p-dioxins on C60 fullerene using Nano-QSPR

Many technological implementations in the field of nanotechnology have involved carbon nanomaterials, including fullerenes such as the buckminsterfullerene, C60. The unprecedented properties of such organic nanomaterials (in particular their large surface area) gained extensive attention for their potential use as organic pollutant sorbents. Sorption interactions can be very hazardous and useful at the same time. This work investigates the influence of halogenation by bromine and/or chlorine in dibenzo-p-dioxins on their sorption ability on the C60 fullerene surface. Halogenated dibenzo-p-dioxins (PXDDs, where X = Br or Cl) are ever-present in the environment and accidently produced in many technological processes in only approximately known quantities. If all combinatorial Br and/or Cl dioxin substitution possibilities are present in the environment, the experimental characterization and investigation of sorbent effectiveness is more than difficult. In this work, we have developed a quantitative structure-property relationship (QSPR) model (R2 = 0.998), predicting the adsorption energy [kcal/mol] for 1,701 PXDDs adsorbed on C60 (PXDD@C60). Based on the QSPR model reported herein, we concluded that the lowest energy PXDD@C60 complexes are those that the World Health Organization (WHO) considers to be less dangerous with respect to the aryl hydrocarbon receptor (AhR) toxicity mechanism. Therefore, the effectiveness of fullerenes as sorbent agents may be underestimated as sorption could be less effective for toxic congeners than previously believed.

Development of polyparameter linear free energy relationship models for octanol-air partition coefficients of diverse chemicals

The octanol-air partition coefficient (K_OA) is a key parameter describing the partition behavior of organic chemicals between air and environmental organic phases. As the experimental determination of K_OA is costly, time-consuming and sometimes limited by the availability of authentic chemical standards for the compounds to be determined, it becomes necessary to develop credible predictive models for K_OA. In this study, a polyparameter linear free energy relationship (pp-LFER) model for predicting K_OA at 298.15 K and a novel model incorporating pp-LFERs with temperature (pp-LFER-T model) were developed from 795 log K_OA values for 367 chemicals at different temperatures (263.15-323.15 K), and were evaluated with the OECD guidelines on QSAR model validation and applicability domain description. Statistical results show that both models are well-fitted, robust and have good predictive capabilities. Particularly, the pp-LFER model shows a strong predictive ability for polyfluoroalkyl substances and organosilicon compounds, and the pp-LFER-T model maintains a high predictive accuracy within a wide temperature range (263.15-323.15 K).

Quantitative modeling of bioconcentration factors of carbonyl herbicides using multivariate image analysis

The bioconcentration factor (BCF) is an important parameter used to estimate the propensity of chemicals to accumulate in aquatic organisms from the ambient environment. While simple regressions for estimating the BCF of chemical compounds from water solubility or the n-octanol/water partition coefficient have been proposed in the literature, these models do not always yield good correlations and more descriptive variables are required for better modeling of BCF data for a given series of organic pollutants, such as some herbicides. Thus, the logBCF values for a set of carbonyl herbicides comprising amide, urea, carbamate and thiocarbamate groups were quantitatively modeled using multivariate image analysis (MIA) descriptors, derived from colored image representations for chemical structures. The logBCF model was calibrated and vigorously validated (r(2) = 0.79, q(2) = 0.70 and rtest(2) = 0.81), providing a comprehensive three-parameter linear equation after variable selection (logBCF = 5.682 - 0.00233 × X9774 - 0.00070 × X813 - 0.00273 × X5144); the variables represent pixel coordinates in the multivariate image. Finally, chemical interpretation of the obtained models in terms of the structural characteristics responsible for the enhanced or reduced logBCF values was performed, providing key leads in the prospective development of more eco-friendly synthetic herbicides.

Predicting the reaction rate constants of micropollutants with hydroxyl radicals in water using QSPR modeling

Quantitative structure-property relationship (QSPR) models which predict hydroxyl radical rate constants (kOH) for a wide range of emerging micropollutants are a cost effective approach to assess the susceptibility of these contaminants to advanced oxidation processes (AOPs). A QSPR model for the prediction of kOH of emerging micropollutants from their physico-chemical properties was developed with special attention to model validation, applicability domain and mechanistic interpretation. In this study, 118 emerging micropollutants including those experimentally determined by the author and data collected from the literature, were randomly divided into the training set (n=89) and validation set (n=29). 951 DRAGON molecular descriptors were calculated for model development. The QSPR model was calibrated by applying forward multiple linear regression to the training set. As a result, 7 DRAGON descriptors were found to be important in predicting the kOH values which related to the electronegativity, polarizability, and double bonds, etc. of the compounds. With outliers identified and removed, the final model fits the training set very well and shows good robustness and internal predictivity. The model was then externally validated with the validation set showing good predictive power. The applicability domain of the model was also assessed using the Williams plot approach. Overall, the developed QSPR model provides a valuable tool for an initial assessment of the susceptibility of micropollutants to AOPs.

QSARINS-chem: Insubria datasets and new QSAR/QSPR models for environmental pollutants in QSARINS

A database of environmentally hazardous chemicals, collected and modeled by QSAR by the Insubria group, is included in the updated version of QSARINS, software recently proposed for the development and validation of QSAR models by the genetic algorithm-ordinary least squares method. In this version, a module, named QSARINS-Chem, includes several datasets of chemical structures and their corresponding endpoints (physicochemical properties and biological activities). The chemicals are accessible in different ways (CAS, SMILES, names and so forth) and their three-dimensional structure can be visualized. Some of the QSAR models, previously published by our group, have been redeveloped using the free online software for molecular descriptor calculation, PaDEL-Descriptor. The new models can be easily applied for future predictions on chemicals without experimental data, also verifying the applicability domain to new chemicals. The QSAR model reporting format (QMRF) of these models is also here downloadable. Additional chemometric analyses can be done by principal component analysis and multicriteria decision making for screening and ranking chemicals to prioritize the most dangerous.

From descriptors to predicted properties: experimental design by using applicability domain estimation

The importance of reliable methods for representative sub-sampling in terms of experimental design and risk assessment within the European Registration, Evaluation, Authorisation and Restriction of Chemicals (REACH) system is crucial. We developed experimental design approaches, by utilising predicted properties and the 'distance to model' parameter, to estimate the benefits of certain compounds to the quality of a resulting model. A statistical evaluation of four regression data sets and one classification data set showed that the adaptive concept of iteratively refining the representation of the chemical space contributes to a more efficient and more reliable selection in comparison to traditional approaches. The evaluation of compounds with regard to the uncertainty and the correlation of prediction is beneficial, and in particular, for regression data sets of sufficient size, whereas the use of predicted properties to define the chemical space is beneficial for classification models.

QSAR Modeling is not "Push a Button and Find a Correlation": A Case Study of Toxicity of (Benzo-)triazoles on Algae

A case study of toxicity of (benzo)triazoles ((B)TAZs) to the algae Pseudokirchneriella subcapitata is used to discuss some problems and solutions in QSAR modeling, particularly in the environmental context. The relevance of data curation (not only of experimental data, but also of chemical structures and input formats for the calculation of molecular descriptors), the crucial points of QSAR model validation and the potential application for new chemicals (internal robustness, exclusion of chance correlation, external predictivity, applicability domain) are described, while developing MLR-OLS models based on molecular descriptors, calculated by various QSAR software tools (commercial DRAGON, free PaDEL-Descriptor and QSPR-THESAURUS). Additionally, the utility of consensus models is highlighted. This work summarizes a methodology for a rigorous statistical approach to obtain reliable QSAR predictions, also for a large number of (B)TAZs in the ECHA preregistration list of REACH (even if starting from limited experimental data availability), and has evidenced some ambiguities and discrepancies related to SMILES notations from different databases; furthermore it highlighted some general problems related to QSAR model generation and was useful in the implementation of the PaDEL-Descriptor software.

QSAR model reproducibility and applicability: a case study of rate constants of hydroxyl radical reaction models applied to polybrominated diphenyl ethers and (benzo-)triazoles

The crucial importance of the three central OECD principles for quantitative structure-activity relationship (QSAR) model validation is highlighted in a case study of tropospheric degradation of volatile organic compounds (VOCs) by OH, applied to two CADASTER chemical classes (PBDEs and (benzo-)triazoles). The application of any QSAR model to chemicals without experimental data largely depends on model reproducibility by the user. The reproducibility of an unambiguous algorithm (OECD Principle 2) is guaranteed by redeveloping MLR models based on both updated version of DRAGON software for molecular descriptors calculation and some freely available online descriptors. The Genetic Algorithm has confirmed its ability to always select the most informative descriptors independently on the input pool of variables. The ability of the GA-selected descriptors to model chemicals not used in model development is verified by three different splittings (random by response, K-ANN and K-means clustering), thus ensuring the external predictivity of the new models, independently of the training/prediction set composition (OECD Principle 5). The relevance of checking the structural applicability domain becomes very evident on comparing the predictions for CADASTER chemicals, using the new models proposed herein, with those obtained by EPI Suite.

QSAR model for the prediction of bio-concentration factor using aqueous solubility and descriptors considering various electronic effects

The in silico modelling of bio-concentration factor (BCF) is of considerable interest in environmental sciences, because it is an accepted indicator for the accumulation potential of chemicals in organisms. Numerous QSAR models have been developed for the BCF, and the majority utilize the octanol/water partition coefficient (log P) to account for the penetration characteristics of the chemicals. The present work used descriptors from a variety of software packages for the development of a multi-linear regression model to estimate BCF. The modelled data set of 473 diverse compounds covers a wide range of log BCF values. In the proposed QSAR model, most of the variation is described by the calculated solubility in water. Other contributing descriptors describe, for instance, hydrophobic surface area, hydrogen bonding and other electronic effects. The model was validated internally by using a variety of statistical approaches. Two external validations were also performed. For the former validation, a subset from the same data source was used. The 2nd external validation was based on an independent data set collected from different resources. All validations showed the consistency of the model. The applicability domain of the model was discussed and described and a thorough outlier analysis was performed.

Influence of the molecular structure and exposure concentration on the uptake and elimination kinetics, bioconcentration, and biotransformation of anionic and nonionic surfactants

In vivo experiments with fish (sole) were performed for pure homologs of linear alkylbenzene sulfonates (LAS) and alcohol ethoxylates (AEO), the most widely used anionic and nonionic surfactants. The surfactant concentration measured in the organism was higher than in the experimental water, and increased with the exposure concentration for both compounds. At the exposure levels tested the bioconcentration factors (BCF) for AEO were one order of magnitude higher than for LAS. Two linear relationships for hydrophobicity were established, one with BCF and the other with uptake rate. The influence of the exposure concentration on the uptake (k(1)) and elimination (k(2)) velocities was researched. The value obtained for k(1) for AEO was higher than for LAS, while k(2) was very similar for both compounds, indicating differences in the incorporation and but not in the depuration rate. For the first time, internal degradation products of LAS were identified and quantified in fish and water and the glucuronic conjugate of AEO was detected in an organism. The predominant biotransformation process for these compounds may be different: the results suggest a phase I biotransformation for LAS and phase II for AEO, due to their different molecular structures.

A QCAR model for predicting antioxidant activity of wild mushrooms

Wild mushrooms have been described as sources of natural antioxidants, particularly phenolic compounds. However, many other compounds present in wild mushrooms can also act as antioxidants (reducers), so whole extracts from a wide range of species need to be examined. To gain further knowledge in this area, the relationship between the antioxidant potential (scavenging effect and reducing power) and chemical composition of twenty three samples from seventeen Portuguese wild mushroom species was investigated. A wide range of analytical parameters reported by our research group (including ash, carbohydrates, proteins, fat, monounsaturated fatty acids, polyunsaturated fatty acids, saturated fatty acids, phenolics, flavonoids, ascorbic acid and beta-carotene) were studied and the data were analysed by partial least squares (PLS) regression analysis to allow correlation of all the parameters. Antioxidant activity correlated well with phenolic and flavonoid contents. A QCAR (Quantitative Composition-Activity Relationships) model was constructed, using the PLS method, and its robustness and predictability were verified by internal and external cross-validation methods. Finally, this model proved to be a useful tool in the prediction of mushrooms' reducing power.

Determination and estimation of partitioning properties for substituted phosphates and thiophosphates

Substituted phenylglyoxylonitrile oximino phosphates and thiophosphates invested as a synergist of molluscicide killing snail eggs have been exposed to the environment with very little attention. The partitioning properties, aqueous solubility (S(w)), n-octanol-water partition coefficient (K(ow)) and soil organic carbon sorption coefficients (K(oc)) of 16 O, O'-dialkyl, O''-(substituted phenylglyoxylonitrile oximino) phosphates and thiophosphates, were determined by the traditional shaking flask method. The parameters of molecular fragment connectivity indices (MFCIs) and linear solvation energy relationships (LSERs) were used as molecular descriptors to establish a series of correlation equations successfully. The obtained correlation equations provided a quantitative method to predict the three partitioning properties for new exploited substituted phosphates and thiophosphates. More valuable, the successful application of MFCIs provides us with a good example and a good idea to improve traditional molecular connectivity indices (MCIs).

A new hybrid system of QSAR models for predicting bioconcentration factors (BCF)

The aim was to develop a reliable and practical quantitative structure-activity relationship (QSAR) model validated by strict conditions for predicting bioconcentration factors (BCF). We built up several QSAR models starting from a large data set of 473 heterogeneous chemicals, based on multiple linear regression (MLR), radial basis function neural network (RBFNN) and support vector machine (SVM) methods. To improve the results, we also applied a hybrid model, which gave better prediction than single models. All models were statistically analysed using strict criteria, including an external test set. The outliers were also examined to understand better in which cases large errors were to be expected and to improve the predictive models. The models offer more robust tools for regulatory purposes, on the basis of the statistical results and the quality check on the input data.

Characterisation of the chemical and biological properties of molecules with QSAR/QSPR and chemical grouping, and its application to a group of alkyl ethers

This study presents a QSAR/QSPR modelling and chemical grouping (read-across) approach to provide information on the biological properties of a group of aliphatic ethers, with accurate biological predictions restricted to those physico-chemical and (eco)toxicological properties where the performance of QSAR/QSPR has been shown to be acceptable. The mathematical methods used ranged from multivariate regression models to PLS (partial least-squares), SVM (support vector machines) and Sammon's mapping. A novel grouping approach, based on a set of key descriptors, has been proposed to give a compact picture of the structural and biological properties of the compounds, and to provide a more mechanistic basis for the interpretations of chemical groups. Besides being a straightforward case study, the paper also exemplifies the capabilities and limitations of the methods in predictive toxicology on a more general level.

A new predictive model for the bioconcentration factors of polychlorinated biphenyls (PCBs) based on the molecular electronegativity distance vector (MEDV)

Polychlorinated biphenyls (PCBs) are some of the most prevalent pollutants in the total environment and receive more and more concerns as a group of ubiquitous potential persistent organic pollutants. Using the variable selection and modeling based on prediction (VSMP), the molecular electronegativity distance vector (MEDV) derived directly from the molecular topological structures was employed to develop a linear model (MI) between the bioconcentration factors (BCF) and two MEDV descriptors of 58 PCBs. The MI model showed a good estimation ability with a correlation coefficient (r) of 0.9605 and a high stability with a leave-one-out cross-validation correlation coefficient (q) of 0.9564. The MEDV-base model (MI) is easier to use than the splinoid poset method reported by Ivanciuc et al. [Ivanciuc, T., Ivanciuc, O., Klein, D.J., 2006. Modeling the bioconcentration factors and bioaccumulation factors of polychlorinated biphenyls with posetic quatitative super-structure/activity relationships (QSSAR). Mol. Divers. 10, 133-145] and gives a better statistics than molecular connectivity index (MCI)-base model developed by Hu et al. [Hu, H.Y., Xu, F.L., Li, B.G., Cao, J., Dawson, R., Tao, S., 2005. Prediction of the bioconcentration factor of PCBs in fish using the molecular connectivity index and fragment constant models. Water Environ. Res. 77, 87-97]. Main structural factors influencing the BCF of PCBs are the substructures expressed by two atomic groups >C= and -CH=. 58 PCBs were divided into an "odd set" and "even set" in order to ensure the predicted potential of the MI for the external samples. It was shown that three models, MI, MO for "odd set", and ME for "even set", can be used to predict the BCF of remaining 152 PCBs in which the experimental BCFs are not available.

QSAR study of selective ligands for the thyroid hormone receptor β

In this paper, an accurate and reliable QSAR model of 87 selective ligands for the thyroid hormone receptor beta 1 (TRbeta1) was developed, based on theoretical molecular descriptors to predict the binding affinity of compounds with receptor. The structural characteristics of compounds were described wholly by a large amount of molecular structural descriptors calculated by DRAGON. Six most relevant structural descriptors to the studied activity were selected as the inputs of QSAR model by a robust optimization algorithm Genetic Algorithm. The built model was fully assessed by various validation methods, including internal and external validation, Y-randomization test, chemical applicability domain, and all the validations indicate that the QSAR model we proposed is robust and satisfactory. Thus, the built QSAR model can be used to fast and accurately predict the binding affinity of compounds (in the defined applicability domain) to TRbeta1. At the same time, the model proposed could also identify and provide some insight into what structural features are related to the biological activity of these compounds and provide some instruction for further designing the new selective ligands for TRbeta1 with high activity.

Linear QSAR regression models for the prediction of bioconcentration factors by physicochemical properties and structural theoretical molecular descriptors

The development of QSAR models useful for the prediction of fish bioconcentration factor (BCF) for a wide range of different chemical classes is crucial for the assessment and prioritisation of potentially persistent bioaccumulative and toxic substances. In this study we present QSAR models for BCF developed on a wide range of chemical structural classes of environmental and toxicological interest (such as dyes and various chlorinated and brominated compounds). The aim is to provide valid QSAR models, statistically validated for predictivity, for the prediction of BCF in general, but also for problematical chemical classes such as highly hydrophobic chemicals. Several descriptors, calculated by different commercially available software packages, have been employed in order to take into account relevant information provided by physicochemical properties (octanol/water partition coefficient and water solubility) and molecular features (structural and quantum-chemical molecular descriptors). The best descriptor subsets for the models were selected using the Genetic Algorithm-Variable Subset Selection strategy (GA-VSS) and calculations were performed by ordinary least squares regression. Starting from a data set of 640 compounds (logK(ow) range from -2.34 to 12.66), we developed linear QSARs, firstly for a data set of 620 compounds (logK(ow) range from -2.34 to 10.35) and secondly specifically for 87 highly hydrophobic chemicals (logK(ow) range from 6.00 to 10.35). All these models have been statistically validated (both internally by cross-validation and bootstrap and externally, by "a priori" splitting of available data by Kohonen Map-ANN in training and prediction sets) and their structural chemical domain has been verified by the leverage approach.

Approaches for externally validated QSAR modelling of Nitrated Polycyclic Aromatic Hydrocarbon mutagenicity

Nitrated Polycyclic Aromatic Hydrocarbons (nitro-PAHs), ubiquitous environmental pollutants, are recognized mutagens and carcinogens. A set of mutagenicity data (TA100) for 48 nitro-PAHs was modeled by the Quantitative Structure-Activity Relationships (QSAR) regression method, and OECD principles for QSAR model validation were applied. The proposed Multiple Linear Regression (MLR) models are based on two topological molecular descriptors. The models were validated for predictivity by both internal and external validation. For the external validation, three different splitting approaches, D-optimal Experimental Design, Self Organizing Maps (SOM) and Random Selection by activity sampling, were applied to the original data set in order to compare these methodologies and to select the best descriptors able to model each prediction set chemicals independently of the splitting method applied. The applicability domain was verified by the leverage approach.