SIMPÓSIOS

1. Simpósio de Quimiometria para QSAR e Química Computacional

PALESTRANTE:
Alan Talevi (Universidad Nacional de La Plata – La Plata - Argentina)

TÍTULO:

Using PPV surfaces to choose score thresholds: successful virtual screening
applications to discover anti-infective agents

Virtual screening involves using computational models or algorithms to screen large chemical libraries and prioritize which compounds will be submitted to experimental testing. Since their introduction in the virtual screening field, Receiver Operating Characteristic (ROC) -derived metrics have been widely used for benchmarking and optimization purposes of computational methods intended for in silico screening applications. Whereas in classification problems the relation between sensitivity and specificity is very informative, a practical concern in virtual screening campaigns is to predict the actual probability that a predicted hit will prove truly active when submitted to experimental testing (in other words, the Positive Predictive Value - PPV). We have explored the use of PPV surfaces derived from simulation of ranking experiments as a complementary tool to ROC curves, for both benchmarking and optimization purposes. The utility of the proposed approach is illustrated by some successful case examples in the field of anti-infectious diseases, including both potential Chagas disease and malaria drugs.

Palestra em
PORTUGUÊS

PALESTRANTE:
Eugene N. Muratov (University of North Carolina – USA)

TÍTULO:

Chembench, Chemotext, and other publicly-accessible cheminformatics portals

The enormous increase in the amount of publicly available chemical genomics data and the growing emphasis on data sharing and open science mandates that cheminformaticians also make their models publicly available for broad use by the scientific community. Chembench is one of the first publicly accessible, integrated cheminformatics Web portals and Chemotext is a publicly available Web server that mines the entire compendium of published literature. Publicly-accessible cheminformatics portals promote the principles of both open science and data and model sharing in the era of Big Data.

Palestra em
PORTUGUÊS

PALESTRANTE:
Marcus Tullius Scotti (UFPB)

TÍTULO:

Using machine learning methods to extract information from natural products databases

The importance of using natural products for medicinal chemistry has demonstrated success in providing scaffolds for new drugs, but also for other applications. Therefore, the organization of data on the structural diversity of secondary metabolites and their distribution is extremely useful for applications in the field of chemistry of natural products, medicinal chemistry, pharmacognosy and ecology among others. In view of these facts, we are focusing efforts on the use of the machine learning approaches to perform chemotaxonomic studies and/or ligand-based  virtual screening (VS) to select potentially active secondary metabolites from an internal dataset, SistematX (sistematx.ufpb.br), a web tool that as provide various information on secondary metabolites such as its structure, its botanical occurrence, geographical distribution, and experimental data, for example, biological activity. 

Palestra em
PORTUGUÊS

PALESTRANTE:
Marcia M. C. Ferreira (PQ) (UNICAMP)

TÍTULO:

QSAR: from 2D to 4D - QSAR methodologies

In QSAR (Quantitative Structure-Activity Relationships), we are interested in building a mathematical relationship linking chemical structure and pharmacological activity of a series of compounds previously tested.

Classical 2D QSAR is a research field in Medicinal Chemistry that uses conformational, physical chemical and structural properties of previously tested potential agents to build multivariate regression that can be helpful in:

●Understanding and explaining the mechanism of action at the molecular level and

●The design and development of new compounds presenting desirable biological properties.

Since the 1980s, new challenges occurred with the appearance of 3D-, 4D-, …, *D-QSAR. The 3D methodology, CoMFA (Comparative Molecular Field Analysis), introduced by Cramer soon become a cornerstone for QSAR studies. In CoMFA, the descriptors (originated from molecular interaction field), are the interaction energy between a probe and all atoms in the molecule.

The 4D-QSAR analysis proposed by Hopfinger and co-workers in 1997, incorporates conformational and alignment freedoms to the development of 3D-QSAR models by performing molecular state ensemble averaging, i.e., the fourth “dimension”.

LQTA-QSAR1 is a new 4D-QSAR method developed in our laboratory. This methodology simultaneously explores the main features of CoMFA and 4D-QSAR paradigms. Our approach uses the GROMACS free package to carry out molecular dynamics simulations and generate a conformational ensemble profile, CEP, for each ligand (instead of one as in CoMFA). The aligned CEP of each ligand is the input to the LQTAgrid module which generates the molecular descriptors. A 3D box of defined size with 1 Å resolution, is used to compute Lennard-Jones and Coulomb interaction energies at each grid point considering different probes. In general, thousands of descriptors are generated. The CDDA digital filter2 approach (also developed in our lab.) is used as an a priori feature selection. QSAR-Modeling3 is the module used for building and validating PLS regression models. Variable selection is performed with the ordered predictor selection, OPS, algorithm4. Models are thoroughly validated5 applying the leave–N–out cross-validation and y-randomization methods. Applications including QSAR studies independent and dependent of the receptor will be presented.

Palestra em
PORTUGUÊS

2. Quimiometria e Imagens Digitais

Palestra em
PORTUGUÊS

PALESTRANTE:
Carolina Santos Silva (UFPE)

As imagens hiperespectrais consistem numa importante técnica de medida analítica que é capaz de fornecer não só informações químicas de uma determinada amostra, mas também a distribuição espacial dos compostos na amostra analisada. As imagens hiperespectrais e digitais têm ganhado aplicações em inúmeras áreas de conhecimento. Ciências farmacêuticas, agronomia, ciências de alimentos, aplicações forenses, médicas, entre outras, estão dentre as áreas que mais empregam esse tipo de análise. Nesse seminário veremos conceitos introdutórios sobre imagens hiperespectrais e digitais e algumas de suas aplicações.

TÍTULO:

Imagens digitais e hiperespectrais: uma introdução com aplicações

Palestra em
PORTUGUÊS

PALESTRANTE:
Mario Cesar Ugulino de Araujo 

Nesta conferência será abordada uma breve revisão dos artigos científicos do grupo de pesquisa sob minha coordenação, envolvendo o uso de imagens digitais e Quimiometria em Química Analítica. Essa combinação imagens digitais/quimiometria foi inicialmente empregada pelo nosso grupo no desenvolvimento de um método baseado em imagens digitais de séston, teste multivariado T2 de Hotelling e QDA para monitorar diferentes lagoas de uma estação de tratamento de águas residuais. Desde então, vários outros métodos foram desenvolvidos e aplicados à análise de amostras de chá, leite, mel, café, biodiesel, própolis, cachaça, hambúrguer e vinhos. Nestes trabalhos foram usados os seguintes dispositivos para aquisição de imagens digitais: webcam, câmera digital, scanner de impressora, o par webcam/LEDs-NIR e smartphone. No tratamento das imagens digitais adquiridas durante análise foram empregados métodos quimiométricos de classificação e de calibração multivariada sem ou com seleção de variáveis via o algoritmo das projeções sucessivas (SPA), tais como: SIMCA, SPA-LDA, MLR, PLS-DA, PCA–LDA, kNN e PLS. Em geral, os métodos desenvolvidos usando imagens digitais e quimiometria se destacam por ser uma ferramenta analítica com abordagem “eco-friendly” para identificar / detectar / classificar amostras e quantificar analitos de uma maneira simples, rápida, barata, eficiente e não destrutiva.

TÍTULO:

Imagens Digitais e Quimiometria

Lucas Logo.png

Organização:

Contato

  (83) 9 9104-4754

  xiw.quimiometria.uepb@gmail.com

  /xiwquimiometria

© 2020 site criado por Lucas Veríssimo.