Effect of the structural factors of organic compounds on the acute toxicity toward Daphnia magna.

The acute toxicity of organic compounds towards Daphina magna was subjected to QSAR analysis. The two-dimensional simplex representation of molecular structure (2D SiRMS) and the support vector machine (SVM), gradient boosting (GBM) methods were used to develop QSAR models. Adequate regression QSAR models were developed for incubation of 24 h. Their interpretation allowed us to quantitatively describe and rank the well-known toxicophores, to refine their molecular surroundings, and to distinguish the structural derivatives of the fragments that significantly contribute to the acute toxicity (LC50) of organic compounds towards D. magna. Based on the results of the interpretation of the regression models, a molecular design (modification) of highly toxic compounds was performed in order to reduce their hazard. In addition, acceptable classification QSAR models were developed to reliably predict the following mode of action (MOA): specific and non-specific toxicity of organic compounds towards D. magna. When interpreting these models, we were able to determine the structural fragments and the physicochemical characteristics of molecules that are responsible for the manifestation of one of the modes of action. The on-line version of the OCHEM expert system (https://ochem.eu), HYBOT descriptors, and the random forest and SVM methods were used for a comparative QSAR investigation.

Contribution assessment of multiparameter optimization descriptors in CNS penetration.

Assessment of the influence of six physicochemical properties used in the multiparameter optimization (MPO) approach for chemical penetration of the blood-brain barrier was carried out by means of application of logistic regression and multiple linear regression, using a data set of 578 diverse chemicals. It was found that use of an aggregation MPO-score descriptor did not give satisfactory results with central nervous system (CNS)/non-CNS classification. Thus an application of the MPO approach for CNS penetration is ambiguous. An alternative to the MPO approach in this work contains detailed (quantitative) structure-activity relationship analysis using a number of methods (linear discriminant analysis, random forest, support vector machine, Gaussian process). Three properties (molecular weight, number of H-bond donors and octanol-water partition coefficient) yielded optimal categorical models with modest statistical parameters (accuracy 0.730-0.765 for CNS/non-CNS classification). The poor statistics of regression models for the common data set suggested the presence of subsets with different mechanisms of penetrations. Based on graphic comparison of experimental and calculated Cu,b values, subset clusters have satisfactory statistics. The regression models obtained allowed the estimation of descriptor contributions in log Cu,b. This means that medicinal chemists now have a simple additive scheme for at least preliminary quantitative assessment of this important pharmacokinetic parameter.

Six global and local QSPR models of aqueous solubility at pH = 7.4 based on structural similarity and physicochemical descriptors.

Aqueous solubility at pH = 7.4 is a very important property for medicinal chemists because this is the pH value of physiological media. The present work describes the application of three different methods (support vector machine (SVM), random forest (RF) and multiple linear regression (MLR)) and three local quantitative structure-property relationship (QSPR) models (regression corrected by nearest neighbours (RCNN), arithmetic mean property (AMP) and local regression property (LoReP)) to construct stable QSPRs with clear mechanistic interpretation. Our data set contained experimental values of aqueous solubility at pH = 7.4 of 387 chemicals (349 in the training set and 38 in the test set including 16 own measurements). The initial descriptor pool contained 210 physicochemical descriptors, calculated from the HYBOT, DRAGON, SYBYL and VolSurf+ programs. Six QSPRs with good statistics based on fundamentals of aqueous solubility and optimization of descriptor space were obtained. Those models have an RMSE close to experimental error (0.70), and are amenable to physical interpretation. The QSPR models developed in this study may be useful for medicinal chemists. Global MLR, RF and SVM models may be valuable for consideration of common factors that influence solubility. The RCNN, AMP and LoReP local models may be helpful for the optimization of aqueous solubility in small sets of related chemicals.

Improved correlation between animal and human potency of non-steroidal anti-inflammatory drugs using quantitative structure-activity relationships (QSARs).

Animal models are known not to predict human responses well, in general. However, we have been able to demonstrate that, for a series of non-steroidal anti-inflammatory drugs that are or were in clinical use, the incorporation of two simple physicochemical properties results in excellent correlations between human and rodent potencies for anti-inflammatory, analgesic and anti-pyretic activities. This has the potential to allow the use of historical data to improve drug development.

Mechanism-Based QSAR Modeling of Skin Sensitization.

Many chemicals can induce skin sensitization, and there is a pressing need for non-animal methods to give a quantitative indication of potency. Using two large published data sets of skin sensitizers, we have allocated each sensitizing chemical to one of 10 mechanistic categories and then developed good QSAR models for the seven categories that have a sufficient number of chemicals to allow modeling. Both internal and external validation checks showed that each model had good predictivity.

QSAR study of some anti-hyperglycaemic sulphonylurea drugs.

Sulphonylureas are widely used anti-hyperglycaemic drugs for the treatment of type 2 diabetes. The only published quantitative structure-activity relationship (QSAR) models for sulphonylurea drugs have been found to be questionable, for a number of reasons. We have re-analysed the human anti-hyperglycaemic potencies, acute mouse intraperitoneal toxicities (LD50) and plasma protein-binding abilities of the 15 drugs using multiple linear regression and obtained good QSAR models for each endpoint. The obtained QSARs all comply well with the Organisation for Economic Co-operation and Development (OECD) Guidelines for the Validation of (Q)SARs. We could not carry out external validation of our models for acute toxicity and plasma protein-binding because of the very small datasets available.

QSPR prediction of physico-chemical properties for REACH.

For registration of a chemical, European Union REACH legislation requires information on the relevant physico-chemical properties of the chemical. Predicted property values can be used when the predictions can be shown to be valid and adequate. The relevant physico-chemical properties that are amenable to prediction are: melting/freezing point, boiling point, relative density, vapour pressure, surface tension, water solubility, n-octanol-water partition coefficient, flash point, flammability, explosive properties, self-ignition temperature, adsorption/desorption, dissociation constant, viscosity, and air-water partition coefficient (Henry's law constant). Published quantitative structure-property relationship (QSPR) methods for all of these properties are discussed, together with relevant property prediction software, as an aid for those wishing to use predicted property values in submissions to the European Chemicals Agency (ECHA).

QSAR modelling of bioconcentration factor using hydrophobicity, hydrogen bonding and topological descriptors.

Bioconcentration factor (BCF) is an important step in the uptake of environmental pollutants in the food chain. It is expensive and time-consuming to measure, so predictive methods are of value. We have used an artificial neural network QSAR approach involving descriptors for hydrophobicity, hydrogen bonding and molecular topology, obtained from commercially available software, to predict the fish BCF values of a diverse data set of 624 chemicals. The training set statistics were: r²= 0.765, q²= 0.763, s = 0.610, and those of the external test set were: r²= 0.739, s = 0.627. The model complies with the OECD Principles for the Validation of (Q)SARs.

In silico prediction of aqueous solubility: the solubility challenge.

The dissolution of a chemical into water is a process fundamental to both chemistry and biology. The persistence of a chemical within the environment and the effects of a chemical within the body are dependent primarily upon aqueous solubility. With the well-documented limitations hindering the accurate experimental determination of aqueous solubility, the utilization of predictive methods have been widely investigated and employed. The setting of a solubility challenge by this journal proved an excellent opportunity to explore several different modeling methods, utilizing a supplied dataset of high-quality aqueous solubility measurements. Four contrasting approaches (simple linear regression, artificial neural networks, category formation, and available in silico models) were utilized within our laboratory and the quality of these predictions was assessed. These were chosen to span the multitude of modeling methods now in use, while also allowing for the evaluation of existing commercial solubility models. The conclusions of this study were surprising, in that a simple linear regression approach proved to be superior over more complex modeling methods. Possible explanations for this observation are discussed and also recommendations are made for future solubility prediction.

How not to develop a quantitative structure-activity or structure-property relationship (QSAR/QSPR).

Although thousands of quantitative structure-activity and structure-property relationships (QSARs/QSPRs) have been published, as well as numerous papers on the correct procedures for QSAR/QSPR analysis, many analyses are still carried out incorrectly, or in a less than satisfactory manner. We have identified 21 types of error that continue to be perpetrated in the QSAR/QSPR literature, and each of these is discussed, with examples (including some of our own). Where appropriate, we make recommendations for avoiding errors and for improving and enhancing QSAR/QSPR analyses.

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.

Henry's law constant of hydrocarbons in air-water system: the cavity ovality effect on the non-electrostatic contribution term of solvation free energy.

In this study, a quantitative structure-property relationship (QSPR) model for the prediction of Henry's law constants of aliphatic hydrocarbons in air-water system has been developed, based on a data-set of 189 compounds. The well-known linear thermodynamic relation between the logarithm of Henry's law constant and solvation free energy has been used for developing the model. It is emphasised that the solvent-accessible surface area (SASA) descriptor is not adequate for predicting the solvation free energy of a wide range of aliphatic hydrocarbons; there are many compounds that have the same solvent-accessible surface area with different solvation free energy. Therefore, we have introduced cavity ovality as a good descriptor of molecular cavity shape factor. The root mean square error (RMSE) of the QSPR regression model based on SASA improves from 0.40 to 0.22 by introducing the cavity ovality descriptor. The QSPR linear ovality model has good statistical parameters (r(2) = 0.90). To emphasise the significant effect of the new descriptor, a non-linear neural network model with only two nodes in the hidden layer was developed, and also yielded a RMSE of 0.22.

QSARS for toxicity to the bacterium Sinorhizobium meliloti.

In the present study, structure-activity relationship (QSAR) models for the prediction of the toxicity to the bacterium Sinorhizobium meliloti have been developed, based on a data set of 140 compounds. The data set is highly heterogeneous both in terms of chemistry and mechanisms of toxic action. For deriving QSARs, chemicals were divided into groups according to mechanism of action and chemical structure. The QSARs derived are considered to be of moderate statistical quality. A baseline effect (relationship between the toxicity and logP), which can be related to non-polar narcosis, was observed. To explain toxicity greater than the baseline toxicity, other structural descriptors were used. The development of models for non-polar and polar narcosis had some success. It appeared that the toxicity of compounds acting by more specific mechanisms of toxic action is difficult to predict. A global QSAR was also developed, which had square of the correlation coefficient r2 = 0.53. A QSAR with reasonable statistical parameters was developed for the aliphatic compounds in the data set (r2 = 0.83). QSARs could not be obtained for the aromatic compounds as a group.

Qsar investigation of a large data set for fish, algae and Daphnia toxicity.

A large data set containing values for fish, algae and Daphnia toxicity for more than 2000 chemicals and mixtures was investigated. The data set was taken from the New Chemicals Data Base of the European Union [hosted by the European Chemicals Bureau, Joint Research Centre, European Commission (http://ecb.jrc.it)]. The data are submitted by industry, according to the requirements of EU Council Directive 67/548/EEC as amended for the seventh time by EU Council Directive 92/32/EEC. The toxicities of neutral chemicals, salts, metal complexes, as well as chemical mixtures were extracted. A baseline effect was demonstrated by chemicals known to act by a narcotic mechanism of action, i.e., a relationship was observed between the toxicity and the logarithm of the octanol-water partition coefficient (log P). However, the prediction of the toxicity of more reactive chemicals was found to require the use of additional descriptors.

Creation of predictive models of aquatic toxicity of environmental pollutants with different mechanisms of action on the basis of molecular similarity and HYBOT descriptors.

Over half of known industrial pollutants have minimal toxic effect, in line with the concept of "baseline toxicity"; such toxicity usually correlates well with lipophilicity. The remainder require additional descriptors in order to model their toxicity by the QSAR approach. Hence, it has not been possible, to date, to develop common stable QSAR models for the toxicity of diverse chemicals with various modes of action on the basis of simple regression relationships. Any new methodology has to take such different modes of action into account. In our work, we used for this purpose an original combination of the similarity concept and physicochemical descriptors calculated by HYBOT, in order to construct stable QSAR models of guppy toxicity. The training set comprised 293 diverse chemicals. Experimental value(s) of one or more nearest related chemicals were used to take structural features and possible modes of toxic action into account. In addition, molecular polarisability and hydrogen bond descriptors for the chemicals of interest and related compounds were used to calculate any additional contribution in toxicity by means of linear regression relationships. Final comparison of calculated and experimental toxicity values gave good results, with standard deviation close to the experimental error.

Improved prediction of fish bioconcentration factor of hydrophobic chemicals.

Using a large heterogeneous data-set of 640 organic chemicals, we have developed predictive Quantitative Structure-Activity Relationship models for fish bioconcentration factor (BCF). For 539 chemicals with a log Kow (octanol-water partition coefficient) range of -2.3 to 6.0, we developed a model with r2 = 0.664 and a standard error of 0.661; the primary descriptor was log Kow, and others were polarisability, number of amino groups, hydrogen bond acceptor ability and a molecular shape factor. For 101 chemicals with a log Kow range of 6.0-12.7, we developed a model with r2 = 0.710 and a standard error of 0.777; the descriptors were aqueous solubility (reflecting the importance of this property in governing uptake from aqueous solution), polarity, polarisability, hydrogen bond donor ability and molecular size. Bearing in mind the very great range of BCF values of highly hydrophobic chemicals, our model offers good predictivity of this important environmental property.

QSAR studies on P-glycoprotein-regulated multidrug resistance and on its reversal by phenothiazines.

Multidrug resistance is brought about largely by membrane transport proteins such as P-glycoprotein (P-gp). We have developed a quantitative structure-activity relationship (QSAR) for P-gp-associated ATPase activity for a diverse set of 22 drugs, and found that such activity is related to substrate molecular size and polarity. We have also developed a QSAR for drug efflux from the blood-brain barrier of another diverse set of 22 drugs, and found that such efflux is a function of drug size and polarisability. Thirdly, we have carried out a QSAR analysis of the ability of 157 phenothiazines and related drugs to reverse multidrug resistance. We were unable to obtain a good QSAR for the whole data-set, but when we divided the data-set into sub-sets of closely related structures, a series of good correlations was obtained, most of which incorporated descriptors that model molecular size and polarity/polarisability. In no instance did we find any evidence that hydrogen bonding or hydrophobicity play a part in multidrug resistance or its reversal, despite that fact that several other workers have reported that these effects appear to be important here.

Study of local anesthetic activity of some derivatives of 3-amino-benzo-[d]-isothiazole.

On the basis of computer prediction of biological activity by PASS and toxicity by DEREK, the most prospective 18 alkylaminoacyl derivatives of 3-amino-benzo-[d]-isothiazole were selected. Their local anesthetic action was assessed using an in vitro preparation of the isolated peroneal nerve of the frog. The local anesthetics action of the compounds was assessed according to the time required for each compound to reduce the amplitude of the evoked compound action potential (CAP). Lidocaine was used as the control compound. The results show that the tested compounds can be divided into three groups: (a) compounds with action similar to lidocaine, (b) compounds with action lower than lidocaine and (c) compounds which block completely the evoked CAP, but after the compound was removed and replaced with normal saline showed no recovery of the potential at all. QSAR studies showed that polarizability, polarity and presence of five-membered rings in molecules have a positive influence on local anesthetic activity, while contributions of aromatic CH and singly bonded nitrogen are negative. Since estimations from PASS probabilities to find local anesthetic activity in the most active compounds were less than 50%, these compounds may be considered as new chemical entities (NCEs).