L’émergence des méthodes de déréplication comme moyen d’identification rapide des substances naturelles connues implique le développement conjoint d’outils informatiques dédiés au traitement et à l’analyse des données spectrales. Dans ce contexte, les travaux présentés dans ce mémoire ont porté sur le développement de deux méthodes in silico de déréplication par résonance magnétique nucléaire (RMN). La première méthode, DerepCrud, permet l’identification des molécules naturelles à partir d’extraits naturels bruts en utilisant des données de RMN du 13C. La méthode permet de comparer des spectres de RMN 1D du 13C issus de l’analyse d’un extrait naturel à ceux des molécules naturelles répertoriées dans une base de données locale afin de pouvoir identifier les composés majoritaires. La deuxième méthode, BCNet, permet d’exploiter les données RMN bidimensionnelles (HMBC et HSQC) pour la déréplication de molécules naturelles. L’algorithme construit un réseau de corrélations HMBC formés par les signaux appartenant aux différentes molécules constituant un extrait puis isole les signaux de chaque molécule grâce à l’utilisation d’algorithmes de détection de communautés. Les molécules sont enfin identifiées en effectuant une recherche dans la base de données des corrélations HMBC. A la fin de la procédure, la présence des molécules identifiées est confirmée par une comparaison de leurs corrélations HSQC théoriques (aussi issues de la base de données) avec les corrélations expérimentales correspondantes afin de renforcer la précision de l’identification. / The emergence of dereplication strategies as a new tool for the rapid identification of the natural products from complex natural extracts has unveiled a great need for cheminformatic tools for the treatment and analysis of the spectral data. The present thesis deals with the development of in silico dereplication methods based on Nuclear Magnetic Resonance (NMR). The first method, DerepCrud, is based on 13C NMR spectroscopy. It identifies the major compounds contained in a crude natural extract without any need for fractionation. The principle of the method is to compare the 13C NMR spectrum of the analyzed mixture to a series of 13C NMR chemical shifts of natural compounds stored in a local database. The second method, BCNet, is designed to exploit the richness of 2D NMR data (HMBC and HSQC) for the dereplication of the natural products. BCNet traces back the network formed by the HMBC correlations of the molecules present in a naturel extract, then isolates the groups of correlations belonging to the individual molecules using a community detection algorithm. The molecules are identified by searching these correlations within a locally constructed database that associates natural product structures and 2D NMR peak positions. Finally, the HSQC correlations of the molecules identified during the previous step are compared to the experimental HSQC correlations of the studied extract in order to increase the quality of identification accuracy.

Com o surgimento da era computacional com especial aplicação em química, as substâncias de origem naturais puderam ter suas informações armazenadas em bancos de dados. Desta forma, surge a oportunidade de se empregar bancos de dados de produtos naturais e de algumas ferramentas de quimioinformática como os estudos de Quantitative Structure-Retention Relationship (QSRR) para acelerar a identificação de substâncias em estudos metabolômicos. Este trabalho propôs o desenvolvimento de três estudos de QSRR, bem como a construção de um banco de dados (AsterDB) com estruturas químicas da família Asteraceae e informações a elas associadas (ex.: ocorrências botânicas e taxonômicas, atividade biológica, informações analíticas etc.) para auxiliar a desreplicação de substâncias em extratos vegetais. O primeiro estudo foi elaborado com 39 lactonas sesquiterpênicas (LST) analisadas em dois diferentes sistemas de solventes (MeOH-H2O 55:45 e MeCN-H2O 35:65), três grupos de descritores estruturais (2D-descr, 3D-1conf e 3D-weigh), dois diferentes conjuntos para treino e teste (26:13 e 29:10), quatro algoritmos para seleção de descritores (best first, linear forward - LFS, greedy stepwise e algoritmo genético - GA), três diferentes tamanhos de modelos (quatro, cinco e seis descritores) e dois métodos de modelagem (mínimos quadrados parciais - PLS e redes neurais artificiais - ANN). O segundo foi desenvolvido com 50 substâncias de diferentes classes químicas com intuito de avaliar as diferenças entre substâncias analisadas individualmente e em mistura em três diferentes equipamentos e dois métodos cromatográficos. O terceiro foi elaborado com 2.635 estruturas químicas com um teste externo comum a todos os modelos (25%, n = 656), três métodos de separação para teste e treino (partição baseada na resposta e baseada nos preditores 2D e 3D), três diferentes tamanhos de modelos selecionados por GA e dois métodos de modelagem (MLR e redes neurais feed-forward com regularização bayesiana - BRNN). O banco de dados AsterDB foi desenvolvido para ser preenchido de forma gradual e atualmente possui cerca de 2.000 estruturas químicas. O primeiro estudo de QSRR gerou bons modelos capazes de estimar o logaritmo do fator de retenção (logk) das LST com P2>0,81 para o sistema MeCN-H2O. O segundo estudo mostrou que não houve diferença estatística entre as substâncias analisadas individualmente e em mistura (p-valor>0,95) e que a correlação entre os dois métodos cromatográficos e equipamentos utilizados foi reprodutível (R>0,95). Estas análises mostraram que foi possível desenvolver modelos de QSRR para um método cromatográfico e equipamento e transpô-los para outro equipamento seguindo o uso de substâncias em comum. O terceiro estudo produziu modelos com boa capacidade de predição (P2>0,81) utilizando alta amplitude de espaço químico e rigor estatístico. Conclui-se que, estas informações podem ser utilizadas como uma plataforma piloto para análises de dados com objetivo de auxiliar na desreplicação de extratos de plantas em estudos metabolômicos / After the emergence of the computing era with special application in chemistry, all substances from natural sources might have their information stored in databases. Therefore, the opportunity arises to employ natural product databases and some chemoinformatic tools such as QSRR studies to speed up the identification of substances from metabolomic studies. This paper proposes the development of three QSRR studies as well as the building of a database (AsterDB) with chemical structures from the Asteraceae family and related information (i.e.: botanical and taxonomic occurrences, biological activity, analytical information, etc.) aiming to assist the dereplication of substances in plant extracts. The first study was carried out with 39 sesquiterpene lactones (STLs) analysed using two different solvent systems (MeOH-H2O 55:45 and MeCN-H2O 35:65), three groups of structural descriptors (2D-descr, 3D-1conf, and 3D-weigh), two different sets for training and testing (26:13 and 29:10), four algorithms for selection of descriptors (best first, LFS, greedy stepwise, and GA), three different model sizes (four, five, and six descriptors) and two modelling methods (PLS and ANN). The second study was developed with 50 compounds of different chemical classification in order to assess the differences between individual and mixed compounds analysed in three different equipments and two chromatographic methods. The third was elaborated with 2,635 chemical structures with a common external test to all models (25%, n = 656), three separation methods for testing- and training-set (based on response and on 2D and 3D predictors partitions), three different sizes of models selected by GA and two modelling methods (MLR and BrNN). The AsterDB database was developed to be populated gradually and currently, it has about 2,000 chemical structures. The first QSRR study generated good models, able to estimate the logarithm of the retention factor (logk) of STLs with P2>0.81 for the MeCN-H2O system. The second study showed that there was no statistical difference between the substances analysed individually and mixed (p-value>0.95) and the correlation between the two chromatographic methods and equipments used was reproducible (R>0.95). These analyses showed that it was possible to develop QSRR models for a chromatographic method and equipment and translate them into other equipment following the use of substances in common. The third study produced models with good predictive capacity (P2>0.81) using a high range of chemical space and statistical accuracy. In conclusion, this information can be used as a pilot platform for data analysis in order to assist in plant dereplication in metabolomics studies

The growth of chemical knowledge available via online databases opens opportunities for new types of chemical research. In particular, by converting the data into a network, graph theoretical approaches can be used to study chemical reactions. In this thesis several research questions from the field of data science and graph theory are re-formulated for the chemistry-specific data. Firstly, the structure of chemical reactions data was studied using graph theory. It was found that the network of reactions obtained from the Reaxys data was scale-free, that on average any two species were separated by six reactions, and that evidence for a hierarchy of nodes existed, most clearly in that the hubs that combine a large share of connections onto them also facilitate a large proportion of routes across the network. The hierarchy was also evidenced in the clustering and degree correlations of nodes. Next, it was investigated whether Reaxys could be mined to construct a network of reactions and use it to plan and evaluate synthesis routes in two case studies. A number of heuristics were developed to find synthesis routes using the network taking chemical structures into account. These routes were fed into a multi-criteria decision making framework scoring the routes along environmental sustainability considerations. The approach was successful in discovering and scoring synthesis route candidates. It was found that Reaxys lacked process data in many instances. To address this a proposal for extension of the RInChI reaction data format was developed. The final question addressed was whether the network could be used to predict future reactions by using Stochastic Block Models. Block model-based link prediction performed impressively, being able to achieve a classification accuracy of close to 95% during time-split validation on historic data, differentiating future reaction discoveries from random data. Next, a set of transformation suggestions was thus evaluated and a framework for analysing these results was presented. Overall, the thesis was able to further the understanding of the network’s topology and to present a framework allowing the mining of Reaxys to plan synthesis routes and target R&D efforts in a specific area to discover new reactions.

Dans le cadre de la recherche pharmaceutique, les propriétés relatives à l’Absorption, la Distribution, le Métabolisme, l’Elimination (ADME) et la Toxicité (Tox) sont cruciales pour le succès des phases cliniques lors de la conception de nouveaux médicaments. Durant ce processus, la chémoinformatique est régulièrement utilisée afin de prédire le profil ADME-Tox des molécules bioactives et d’améliorer leurs propriétés pharmacocinétiques. Ces modèles de prédiction, basés sur la quantification des relations structure-activité (QSAR), ne sont pas toujours efficaces à cause du faible nombre de données ADME-Tox disponibles et de leur hétérogénéité induite par des différences dans les protocoles expérimentaux, ou encore de certaines erreurs expérimentales. Au cours de cette thèse, nous avons d’abord constitué une base de données contenant 150 000 mesures pour une cinquantaine de propriétés ADME-Tox. Afin de valoriser l’ensemble de ces données, nous avons dans un deuxième temps proposé une plateforme automatique de création de modèles de prédiction QSAR. Cette plateforme, nommée MetaPredict, a été conçue afin d’optimiser chacune des étapes de création d’un modèle statistique, dans le but d’améliorer leur qualité et leur robustesse. Nous avons dans un troisième temps valorisé les modèles obtenus grâce à la plateforme MetaPredict en proposant une application en ligne. Cette application a été développée pour faciliter l’utilisation des modèles, apporter une interprétation simplifiée des résultats et moduler les observations obtenues en fonction des spécificités d’un projet de recherche. Finalement, MetaPredict permet de rendre les modèles ADME-Tox accessibles à l’ensemble des chercheurs. / Absorption, Distribution, Metabolism, Elimination (ADME) and Toxicity (Tox) properties are crucial for the success of clinical trials of a drug candidate. During this process, chemoinformatics is regularly used to predict the ADME-Tox profile of bioactive compounds and to improve their pharmacokinetic properties. In silico approaches have already been developed to improve poor pharmacokinetics and toxicity of lead compounds. These predictive models, based on the quantification of structure-activity relationships (QSAR), were not always efficient enough due to the low number of accessible biological data and their heterogeneity induced by the differences in experimental assays or the significant experimental error. In this thesis, we first built a database containing 150,000 data points for about 50 ADME-Tox properties. In order to valorize all this data, we then proposed an automatic platform for creating predictive models. This platform, called MetaPredict, has been designed to optimize each step of model development, in order to improve their quality and robustness. Third,, we promoted the statistical models using the online application of MetaPredict platform. This application has been developed to facilitate the use of newly built models, to provide a simplified interpretation of the results and to modulate the obtained observations according to the needs of the researchers. Finally, this platform provides an easy access to the ADME-Tox models for the scientific community.

Généralement les modèles QSPR ne sont utilisés que pour prédire des propriétés des corps purs. Dans cette thèse nous avons développé une approche QSPR permettant de prédire des propriétés non additives de mélanges binaires, plus précisément leur caractère azéotropique/zéotropique. Pour parvenir à ce résultat, plusieurs types de modèles quantitatifs et qualitatifs ont été développés. L’approche est originale pour deux raisons. Premièrement, peu de travaux de recherche ont été publiés sur des mélanges dont les propriétés sont non-additives. Deuxièmement, plusieurs nouveaux aspects méthodologiques ont été introduits dans ce travail. Tout d'abord des descripteurs "spéciaux", capables de décrire des mélanges ont été proposés. De plus, un protocole robuste d'obtention et de validation des modèles a été utilisé, et un domaine d'applicabilité des modèles fiable a été proposé. La méthodologie développée pendant cette thèse démontre la fiabilité d'un nouveau concept – les modèles QSPR pour les mélanges. Elle est comparable à d'autres méthodes classiques, quoique n'utilisant qu'un faible nombre de données en comparaison. / Generally, QSPR models are limited to individual compounds. In this thesis we have developed a QSPR approach to predict non-additive properties of binary mixtures, more explicitly their azeotropic behavior. To achieve this, several types of quantitative and qualitative models have been developed. This approach is original for two reasons. First, little research has been published on mixtures whose properties are no additive. Second, several new methodological aspects have been introduced in this work. First of all "special" descriptors able to describe mixtures have been proposed. In addition, a robust protocol for obtaining and validating models was used, and a reliable models applicability domain was proposed. The methodology developed during this thesis demonstrates the consistency of a new concept - the QSPR models for mixtures. It is comparable to other conventional methods, though using only limited data.

Prédire à l’avance quels composés seront toxiques chez l’homme ou non représente un réel challenge dans le monde pharmaceutique. En effet, les mécanismes à l’origine de la toxicité ne sont pas toujours bien connus, et à cela s’ajoute le fait qu’un composé peut devenir néfaste seulement après qu’il ait été métabolisé. Nous proposons ici une approche originale utilisant les graphes condensés de réactions afin de modéliser les réactions métaboliques et prédire le devenir des xénobiotiques dans l’organisme humain. Différentes formes de toxicité sont aussi prédites : la mutagénicité et l’hépatotoxicité. Pour cette seconde toxicité, l’approche utilisée est la première à notre connaissance à prédire avec succès les molécules toxiques décrites par des données autres que résultant d’observations in vivo. / Predict in advance which compounds will be toxic in humans or not is a real challenge in the pharmaceutical world. Indeed, the mechanisms responsible for toxicity are not always well known, and in some case a compound become toxic only after it has been metabolized. We propose here a novel approach using condensed graphs of reactions to model and predict the metabolic fate of xenobiotics in the human body. Various forms of toxicity are also predicted : mutagenicity and hepatotoxicity. For this second toxicity, the approach proposed is the first to our knowledge to successfully predict the toxic molecules described by data other than resulting from observations in vivo.

Submitted by Luciana Ferreira (lucgeral@gmail.com) on 2017-12-18T13:15:58Z No. of bitstreams: 2 Tese - Lívia do Carmo Silva - 2017.pdf: 33507637 bytes, checksum: 4e4e645677db485bad788889218760cd (MD5) license_rdf: 0 bytes, checksum: d41d8cd98f00b204e9800998ecf8427e (MD5) / Approved for entry into archive by Luciana Ferreira (lucgeral@gmail.com) on 2017-12-18T13:17:23Z (GMT) No. of bitstreams: 2 Tese - Lívia do Carmo Silva - 2017.pdf: 33507637 bytes, checksum: 4e4e645677db485bad788889218760cd (MD5) license_rdf: 0 bytes, checksum: d41d8cd98f00b204e9800998ecf8427e (MD5) / Made available in DSpace on 2017-12-18T13:17:23Z (GMT). No. of bitstreams: 2 Tese - Lívia do Carmo Silva - 2017.pdf: 33507637 bytes, checksum: 4e4e645677db485bad788889218760cd (MD5) license_rdf: 0 bytes, checksum: d41d8cd98f00b204e9800998ecf8427e (MD5) Previous issue date: 2017-12-01 / Coordenação de Aperfeiçoamento de Pessoal de Nível Superior - CAPES / Paracoccidioidomycosis (PCM) is the cause of several deaths from systemic mycoses. The etiological agents of PCM belong to the genus Paracoccidioides spp., restricted to the regions of Latin America. The infection is acquired by inhaling conidia that primarily settle in the lungs, and can spread to other organs. The treatment of PCM is commonly performed with administration of antifungals such as amphotericin B, itraconazole and co-trimoxazole. The toxicity and side effects of antifungals, added over the long treatment time, has stimulated research for new bioactive compounds. Thus, with the objective of to identify the anti- Paracoccidioides spp. of compounds derived from chalcones and nitrogen heterocycles and to identify the molecular targets and mode of action of argentilactone and RRF-128 in P. brasiliensis were used methodologies such as shape-based virtual screening, minimum inhibitory and fungicidal concentration, cytotoxicity in fibroblast cells, interactions between antifungal, proteomic and chemoproteomics. After the virtual screening, 33 chalcones were proposed as anti-Paracoccidioides molecules, being this activity confirmed by biological assays. Among the compounds, eight aryl and heteroaryl chalcones had selectivity index considered attractive, highlighting Labmol-75 with selectivity index of 64.4 in P. lutzii and 32.2 in P. brasiliensis. In addition, Labmol-75 showed additive interaction with amphotericin B and co-trimoxazole. In relation to nitrogen heterocycles, of the 22 tested compounds, RRF-128 was the most important. RRF-128 showed to inhibit the growth of P. brasiliensis in low concentrations, selectivity index of 64.10, interacting synergistically with itraconazole. In addition, the proteomic analyzes of P. brasiliensis in the presence of the compound provided evidence that the energy metabolism of the fungus is induced to produce acetyl-CoA and that the synthesis of membrane sterols is impaired. In relation to argentilactone, 331 proteins were identified as ligands to this compound in the chemoproteomics assay and after being functionally classified, it was observed that the most representative functional classes are related to amino acid metabolism, energetic and detoxification. The inhibition of the enzymatic activity of malate dehydrogenase, citrate synthase and pyruvate dehydrogenase by argentilactone was confirmed. In addition, argentilactone induced the production of reactive oxygen species, and inhibited chitin and glucan synthesis and arrest of the cell cycle in the G0/G1 phase. From these results, it can be concluded that the compounds Labmol-75, RRF- 128 and argentilactone showed to be promising antifungal agents. / Paracoccidioidomicose (PCM) é a causa de várias mortes por micoses sistêmicas. Os agentes etiológicos da PCM pertencem ao gênero Paracoccidioides spp., restritos às regiões da América Latina. A infecção é adquirida por inalação de conídios que primariamente se instalam nos pulmões, podendo disseminar para outros órgãos. O tratamento da PCM é comumente realizado com a administração de antifúngicos como anfotericina B, itraconazol e co-trimoxazol. A toxidade e efeitos colaterais dos antifúngicos, adicionado ao longo tempo de tratamento, têm impulsionado pesquisas por novos compostos bioativos. Assim, com objetivo de identificar a atividade anti-Paracoccidioides spp. de compostos derivados de chalconas e heterociclos nitrogenados e identificar os alvos moleculares e modo de ação de argentilactona e RRF-128 em P. brasiliensis foram empregadas metodologias como rastreio virtual shape-based, concentração inibitória e fungicida mínima, citotoxicidade em células de fibroblastos, interações entre antifúngicos, proteômica e quimioproteômica. Após o rastreio virtual, 33 chalconas foram propostas como moléculas anti-Paracoccidioides, sendo esta atividade confirmada pelos ensaios biológicos. Dentre os compostos, oito aril e heteroaril chalconas tiveram índices de seletividades considerados promissores, destacando-se Labmol-75 com índice de seletividade de 64,4 em P. lutzii e 32,2 em P. brasiliensis. Além disso, Labmol-75 apresentou interação aditiva com anfotericina B e co-trimoxazol. Dos 22 compostos heterociclos nitrogenados, o RRF-128 apresentou resultados promissor. RRF-128 foi capaz de inibir o crescimento de P. brasiliensis em baixas concentrações e apresentou índice de seletividade de 64,10. Além disso, RRF-128 interagiu de forma sinérgica com itraconazol. As análises proteômicas de P. brasiliensis na presença de RRF-128 forneceram evidências de que o metabolismo energético do fungo é direcionado para produção de acetil-CoA e que a síntese de esteróis de membrana está comprometida. Quanto à argentilactona, 331 proteínas foram identificadas como ligantes a este composto utilizando abordagem quimioproteômica e após serem classificadas funcionalmente, observou-se que as classes funcionais mais representativas são relacionadas ao metabolismo de aminoácidos, energético e detoxificação. A inibição da atividade enzimática de malato desidrogenase, citrato sintase e piruvato desidrogenase por argentilactona foi confirmada. Além disso, argentilactona induziu a produção de espécies reativas de oxigênio, inibiu síntese de quitina e glicana, bem como o aprisionamento do ciclo celular na fase G0/G1. A partir destes resultados, conclui-se que os compostos Labmol-75, RRF-128 e argentilactona são promissores antifúngicos.

Submitted by Erika Demachki (erikademachki@gmail.com) on 2016-08-23T18:34:08Z No. of bitstreams: 2 Dissertação_José Teófilo Moreira Filho.pdf: 3739532 bytes, checksum: 781ef593c7c871e1b885d3e654832a3c (MD5) license_rdf: 0 bytes, checksum: d41d8cd98f00b204e9800998ecf8427e (MD5) / Approved for entry into archive by Erika Demachki (erikademachki@gmail.com) on 2016-08-23T18:34:31Z (GMT) No. of bitstreams: 2 Dissertação_José Teófilo Moreira Filho.pdf: 3739532 bytes, checksum: 781ef593c7c871e1b885d3e654832a3c (MD5) license_rdf: 0 bytes, checksum: d41d8cd98f00b204e9800998ecf8427e (MD5) / Made available in DSpace on 2016-08-23T18:34:31Z (GMT). No. of bitstreams: 2 Dissertação_José Teófilo Moreira Filho.pdf: 3739532 bytes, checksum: 781ef593c7c871e1b885d3e654832a3c (MD5) license_rdf: 0 bytes, checksum: d41d8cd98f00b204e9800998ecf8427e (MD5) Previous issue date: 2016-07-29 / Conselho Nacional de Pesquisa e Desenvolvimento Científico e Tecnológico - CNPq / Schistosomiasis is a neglected tropical disease caused by parasites of the genus Schistosoma. Worldwide, there are about 240 million people infected and have more than 700 million people at risk of infection in 78 countries. Biomphalaria glabrata is the main intermediate host of Schistosoma mansoni in Brazil. Niclosamide is the molluscicide recommended by the WHO. However, this molluscicide is toxic to other species of aquatic animals and plants, has difficult solubilization both in organic solvents and in water and also high cost. The utilization of in silico methods for virtual screening of new compounds rationalizes costs, reduces the time and also the number of animals in the early stages of research and development. The work’s purpose was plan and identify new molluscicidal compounds potentially active against B. glabrata through in silico methods. Known active and inactive molluscicidal compounds against B. glabrata were selected from literature. Decoys were generated to validate the shape-based models and QSAR models. First, the top 10,000 compounds from Chembridge and ZINC "DrugsNow" databases were screened with the best shape-based model, selected using TanimotoCombo socre function. Later, the activity of the compounds was predicted using consensus QSAR models. Finally, the water solubility of the compounds was calculated for the best molluscicidal candidates, leading to the identification of 20 potentially active compounds as molluscicides against B. glabrata. / A esquistossomose é uma doença tropical negligenciada causada por parasitos do gênero Schistosoma. No mundo, existem cerca de 240 milhões de pessoas infectadas e mais de 700 milhões de pessoas em 78 países sob risco de infecção. Biomphalaria glabrata é o principal hospedeiro intermediário de Schistosoma mansoni no Brasil. A niclosamida é o moluscicida recomendado pela Organização Mundial de Saúde. No entanto, este moluscicida é tóxico a outras espécies de animais aquáticos e plantas, possui difícil solubilização tanto em solventes orgânicos quanto em água e, também, possui alto custo. A utilização de métodos in silico para triagem virtual de novos compostos racionaliza os custos, diminui o tempo e reduz número de animais nas fases iniciais de pesquisas e desenvolvimento. O objetivo deste trabalho foi planejar e identificar novos compostos moluscicidas potencialmente ativos contra B. glabrata através de métodos in silico. Compostos ativos e inativos como moluscicidas contra B. glabrata foram selecionados na literatura. Decoys foram gerados para a validação dos modelos de similaridade pela forma 3D e modelos de QSAR. Primeiramente, os melhores 10.000 compostos das bases de dados Chembridge e ZINC "DrugsNow" foram triados com o melhor modelo de similaridade pela forma 3D, utilizando a função de score TanimotoCombo. Posteriormente, a atividade dos compostos foi predita utilizando os modelos de QSAR gerados. Ao final, a solubilidade em água dos compostos com melhor atividade predita foi calculada, levando à identificação de 20 compostos potencialmente moluscicidas contra B. glabrata.

Computational drug repurposing aims at finding new medical uses for existing drugs. The identification of novel drug-target interactions (DTIs) can be a useful part of such a task. Computational determination of DTIs is a convenient strategy for systematic screening of a large number of drugs in the attempt to identify new DTIs at low cost and with reasonable accuracy. This necessitates development of accurate computational methods that can help focus on the follow-up experimental validation on a smaller number of highly likely targets for a drug. Although many methods have been proposed for computational DTI prediction, they suffer the high false positive prediction rate or they do not predict the effect that drugs exert on targets in DTIs. In this report, first, we present a comprehensive review of the recent progress in the field of DTI prediction from data-centric and algorithm-centric perspectives. The aim is to provide a comprehensive review of computational methods for identifying DTIs, which could help in constructing more reliable methods. Then, we present DDR, an efficient method to predict the existence of DTIs. DDR achieves significantly more accurate results compared to the other state-of-theart methods. As supported by independent evidences, we verified as correct 22 out of the top 25 DDR DTIs predictions. This validation proves the practical utility of DDR, suggesting that DDR can be used as an efficient method to identify 5 correct DTIs. Finally, we present DDR-FE method that predicts the effect types of a drug on its target. On different representative datasets, under various test setups, and using different performance measures, we show that DDR-FE achieves extremely good performance. Using blind test data, we verified as correct 2,300 out of 3,076 DTIs effects predicted by DDR-FE. This suggests that DDR-FE can be used as an efficient method to identify correct effects of a drug on its target.

No description available.