Academic literature on the topic 'QSAR Model'

Un modèle statistique de relations structure-activité de type quantitatif des cytokinines, hormones végétales, a été mis au point. Il a été construit à partir de 26 molécules appartenant aux deux principales classes de cytokinines : les pyridylphenylurées et les purines n#6-substituées. Apres l'analyse conformationnelle de ces 26 molécules, différents alignements ont été construits et testes dans COMFA. Finalement, deux modèles ont été sélectionnés. Ils ont été valides par des échantillons-tests de molécules connues. Le meilleur modèle, ORL1, a ensuite servi pour faire des prédictions d'activité sur des molécules hypothétiques. Ces molécules ont par la suite été synthétisées et leur activité biologique a été déterminée. Il s'agit d'une guanidine prédite active mais qui s'est révélée instable après des examens complémentaires. Deux pyridylphenylurées portant un groupement sulfinyl ou sulfonyl en 6 sur la pyridine se sont révélées actives, comme le prédisait le modèle. D'autre part, deux pyridylbenzylurées, analogues des deux molécules précédentes, prédites actives par le modèle, possèdent en fait un effet inhibiteur. Le modèle choisi (ORL1) semble fiable pour la prédiction de molécules rigides telles que les pyridylphenylurées ou les styrylpurines.

The thesis aimed to extract information relevant to the hazard and risk assessment of pesticides. In particular, quantitative structure-activity relationship (QSAR) approaches have been used to build up a mathematical model able to predict the aquatic acute toxicity, Leso, and the avian oral toxicity, LDso, for pesticides. Ecotoxicological values were collected from several databases, and screened according to quality criteria. A hierarchical QSAR approach was applied for the prediction of acute aquatic toxicity. Chemical structures were encoded into molecular descriptors by an automated, seamless procedure available within the OpenMolGRID system. Different linear and non-linear regression techniques were used to obtain reliable and thoroughly validated QSARs. The final model was developed by a counter-propagation neural network coupled with genetic algorithms for variable selection. The proposed QSAR is consistent with McFarland's principle for biological activity and makes use of seven molecular descriptors. The model was assessed thoroughly in test (R2 = 0.8) and validation sets (R2 = 0.72), the y-scrambling test and a sensitivity/stability test. The second endpoint considered in this thesis was avian oral toxicity. As previously, the chemical description of chemicals was generated automatically by the OpenMolGRID system. The best classification model was chosen on the basis of the performances on a validation set of 19 data points, and was obtained from a support vector machine using 94 data points and nine variables selected by genetic algorithms (Error Ratetraining = 0.021, Error Ratevaiidalion = 0.158). The model allowed for a mechanistic estimation of the toxicological action. In fact, several descriptors selected for the final classification model encode for the interaction of the pesticides with other molecules. The presence of hetero-atoms, e.g. sulphur atoms, is correlated with the toxicity, and the pool of descriptor selected is generally dependent from the 3D conformation of the structures. These suggest that, in the case of avian oral toxicity, pesticides probably exert their toxic action through the interaction with some macromolecule and/or protein of the biological system.

Toxic compounds, such as pesticides, are routinely tested against a range of aquatic, avian and mammalian species as part of the registration process. The need for reducing dependence on animal testing has led to an increasing interest in alternative methods such as in silico modelling. The QSAR (Quantitative Structure Activity Relationship)-based models are already in use for predicting physicochemical properties, environmental fate, eco-toxicological effects, and specific biological endpoints for a wide range of chemicals. Data plays an important role in modelling QSARs and also in result analysis for toxicity testing processes. This research addresses number of issues in predictive toxicology. One issue is the problem of data quality. Although large amount of toxicity data is available from online sources, this data may contain some unreliable samples and may be defined as of low quality. Its presentation also might not be consistent throughout different sources and that makes the access, interpretation and comparison of the information difficult. To address this issue we started with detailed investigation and experimental work on DEMETRA data. The DEMETRA datasets have been produced by the EC-funded project DEMETRA. Based on the investigation, experiments and the results obtained, the author identified a number of data quality criteria in order to provide a solution for data evaluation in toxicology domain. An algorithm has also been proposed to assess data quality before modelling. Another issue considered in the thesis was the missing values in datasets for toxicology domain. Least Square Method for a paired dataset and Serial Correlation for single version dataset provided the solution for the problem in two different situations. A procedural algorithm using these two methods has been proposed in order to overcome the problem of missing values. Another issue we paid attention to in this thesis was modelling of multi-class data sets in which the severe imbalance class samples distribution exists. The imbalanced data affect the performance of classifiers during the classification process. We have shown that as long as we understand how class members are constructed in dimensional space in each cluster we can reform the distribution and provide more knowledge domain for the classifier.

As ferramentas de modelagem molecular e de estudos das relações quantitativas entre a estrutura e atividade (QSAR) ou estrutura e propriedade (QSPR) estão integradas ao processo de planejamento de fármacos, sendo de extremo valor na busca por novas moléculas bioativas com propriedades farmacocinéticas e farmacodinâmicas otimizadas. O trabalho em Química Medicinal realizado nesta dissertação de mestrado teve como objetivo estudar as relações quantitativas entre a estrutura e as propriedades farmacocinéticas biodisponibilidade oral e ligação às proteínas plasmáticas. Para a realização deste trabalho, conjuntos padrões de dados foram organizados para as propriedades biodisponibilidade e ligação às proteínas plasmáticas contendo a informação qualificada sobre a estrutura química e a propriedade alvo correspondente. Os conjuntos de dados criados formaram as bases científicas para o desenvolvimento dos modelos preditivos empregando os métodos holograma QSAR e VolSurf. Os modelos finais de HQSAR e VolSurf gerados neste trabalho possuem elevada consistência interna e externa, apresentando bom poder de correlação e predição das propriedades alvo. Devido à simplicidade, robustez e consistência, estes modelos são guias úteis em Química Medicinal nos estágios iniciais do processo de descoberta e desenvolvimento de fármacos.<br>Molecular modeling tools and quantitative structure-activity relantionships (QSAR) or structure-property (QSPR) are integrated into the drug design process in the search for new bioactive molecules with good pharmacokinetic and pharmacodynamic properties. The Medicinal Chemistry work carried out in this Master’s dissertation concerned studies of the quantitative relationshisps between chemical structure and the pharmacokinetic properties oral bioavailability and plasma protein binding. In the present work, standard data sets for bioavailability and plasma protein binding were organized encompassing the structural information and corresponding pharmacokinetic data. The created data sets established the scientific basis for the development of predictive models using the hologram QSAR and VolSurf methods. The final HQSAR and VolSurf models posses high internal and external consistency with good correlative and predictive power. Due to the simplicity, robustness and effectivess, these models are useful guides in Medicinal Chemistry in the early stages of the drug discovery and development process.

Organic pollutants that resist degradation in the environment can accumulate in body tissues and cause unavoidable intoxications to organisms in wild life as well as humans. The possible effects, usually increasing with the cumulative exposure to such chemicals, are not always addressed adequately in risk assessment procedures evaluating long and short-term contact hazard. Thus, chemicals accumulation, degradation and environmental fate are of prime concern for REACH when defining side effects due to chronic exposure. Characteristics and behavior of organic pollutants have been investigated experimentally during the last decades by use of various methods of trace analysis. However, the available data still contains several gaps. In this aim, REACH promotes the use of alternative methods to reduce the number of animal tests and suggests in-silico methods such as Quantitative Structure-Activity Relationships (QSARs) to fill the lack of knowledge. The goal of this thesis, in the framework of the ECO-ITN project, was to build QSAR models with high reliability based on good experimental data for optimal estimation of environmental endpoints of interest for REACH. New molecular descriptors and feature selection techniques have been tested paying particular attention to the validation steps and applicability domain definition.

With the recent advantages of machine learning in cheminformatics, the drug discovery process has been accelerated; providing a high impact in the field of medicine and public health. Molecular property and activity prediction are key elements in the early stages of drug discovery by helping prioritize the experiments and reduce the experimental work. In this thesis, a novel approach for multi-task regression using a text-based Transformer model is introduced and thoroughly explored for training on a number of properties or activities simultaneously. This multi-task regression with Transformer based model is inspired by the field of Natural Language Processing (NLP) which uses prefix tokens to distinguish between each task. In order to investigate our architecture two data categories are used; 133 biological activities from ExCAPE database and three physical chemistry properties from MoleculeNet benchmark datasets. The Transformer model consists of the embedding layer with positional encoding, a number of encoder layers, and a Feedforward Neural Network (FNN) to turn it into a regression problem. The molecules are represented as a string of characters using the Simplified Molecular-Input Line-Entry System (SMILES) which is a ’chemistry language’ with its own syntax. In addition, the effect of Transfer Learning is explored by experimenting with two pretrained Transformer models, pretrained on 1.5 million and on 100 million molecules. The text-base Transformer models are compared with a feature-based Support Vector Regression (SVR) with the Tanimoto kernel where the input molecules are encoded as Extended Connectivity Fingerprint (ECFP), which are calculated features. The results have shown that Transfer Learning is crucial for improving the performance on both property and activity predictions. On bioactivity tasks, the larger pretrained Transformer on 100 million molecules achieved comparable performance to the feature-based SVR model; however, overall SVR performed better on the majority of the bioactivity tasks. On the other hand, on physicochemistry property tasks, the larger pretrained Transformer outperformed SVR on all three tasks. Concluding, the multi-task regression architecture with the prefix token had comparable performance with the traditional feature-based approach on predicting different molecular properties or activities. Lastly, using the larger pretrained models trained on a wide chemical space can play a key role in improving the performance of Transformer models on these tasks.

<p>Rational conversion of biologically active peptides to nonpeptide compounds with retained activity is an appealing approach in drug development. One important objective of the work presented in this thesis was to use computational modeling to aid in such a conversion of the peptide angiotensin II (Ang II, Asp-Arg-Val-Tyr-Ile-His-Pro-Phe). An equally important objective was to gain an understanding of the requirements for ligand binding to the Ang II receptors, with a focus on interactions with the AT₂ receptor.</p><p>The bioactive conformation of a peptide can provide important guidance in peptidomimetic design. By designing and introducing well-defined secondary structure mimetics into Ang II the bioactive conformation can be addressed. In this work, both γ- and β-turn mimetic scaffolds have been designed and characterized for incorporation into Ang II. Using conformational analysis and the pharmacophore recognition method DISCO, a model was derived of the binding mode of the pseudopeptide Ang II analogues. This model indicated that the positioning of the Arg side chain was important for AT₂ receptor binding, which was also supported when the structure–activity relationship of Ang II was investigated by performing a glycine scan.</p><p>To further examine ligand binding, a 3D model of the AT₂ receptor was constructed employing homology modeling. Using this receptor model in a docking study of the ligands, binding modes were identified that were in agreement with data from point-mutation studies of the AT₂ receptor.</p><p>By investigating truncated Ang II analogues, small pseudopeptides were developed that were structurally similar to nonpeptide AT₂ receptor ligands. For further guidance in ligand design of nonpeptide compounds, three-dimensional quantitative structure–activity relationship models for AT₁ and AT₂ receptor affinity as well as selectivity were derived. </p>

Rational conversion of biologically active peptides to nonpeptide compounds with retained activity is an appealing approach in drug development. One important objective of the work presented in this thesis was to use computational modeling to aid in such a conversion of the peptide angiotensin II (Ang II, Asp-Arg-Val-Tyr-Ile-His-Pro-Phe). An equally important objective was to gain an understanding of the requirements for ligand binding to the Ang II receptors, with a focus on interactions with the AT2 receptor. The bioactive conformation of a peptide can provide important guidance in peptidomimetic design. By designing and introducing well-defined secondary structure mimetics into Ang II the bioactive conformation can be addressed. In this work, both γ- and β-turn mimetic scaffolds have been designed and characterized for incorporation into Ang II. Using conformational analysis and the pharmacophore recognition method DISCO, a model was derived of the binding mode of the pseudopeptide Ang II analogues. This model indicated that the positioning of the Arg side chain was important for AT2 receptor binding, which was also supported when the structure–activity relationship of Ang II was investigated by performing a glycine scan. To further examine ligand binding, a 3D model of the AT2 receptor was constructed employing homology modeling. Using this receptor model in a docking study of the ligands, binding modes were identified that were in agreement with data from point-mutation studies of the AT2 receptor. By investigating truncated Ang II analogues, small pseudopeptides were developed that were structurally similar to nonpeptide AT2 receptor ligands. For further guidance in ligand design of nonpeptide compounds, three-dimensional quantitative structure–activity relationship models for AT1 and AT2 receptor affinity as well as selectivity were derived.