Search

The grading of gemstones is currently a manual procedure performed by gemologists. A popular approach uses reference stones, where those are visually inspected by specialists that decide which one of the available reference stone is the most similar to the inspected stone. This procedure is very subjective as different specialists may end up with different grading choices. This work proposes a complete framework that entails the image acquisition and goes up to the final stone categorization. The proposal is able to automate the entire process apart from including the stone in the created chamber for the image acquisition. It discards the subjective decisions made by specialists. This is the first work to propose a machine learning approach coupled with image processing techniques for emerald grading. The proposed framework achieves 98% of accuracy (correctly categorized stones), outperforming a deep learning approach. Furthermore, we also create and publish the used dataset that contains 192 images of emerald stones along with their extracted and pre-processed features.

Background Interactions between proteins and non-proteic small molecule ligands play important roles in the biological processes of living systems. Thus, the development of computational methods to support our understanding of the ligand-receptor recognition process is of fundamental importance since these methods are a major step towards ligand prediction, target identification, lead discovery, and more. This article presents visGReMLIN, a web server that couples a graph mining-based strategy to detect motifs at the protein-ligand interface with an interactive platform to visually explore and interpret these motifs in the context of protein-ligand interfaces. Results To illustrate the potential of visGReMLIN, we conducted two cases in which our strategy was compared with previous experimentally and computationally determined results. visGReMLIN allowed us to detect patterns previously documented in the literature in a totally visual manner. In addition, we found some motifs that we believe are relevant to protein-ligand interactions in the analyzed datasets. Conclusions We aimed to build a visual analytics-oriented web server to detect and visualize common motifs at the protein-ligand interface. visGReMLIN motifs can support users in gaining insights on the key atoms/residues responsible for protein-ligand interactions in a dataset of complexes.

Motivation The discovery of protein–ligand-binding sites is a major step for elucidating protein function and for investigating new functional roles. Detecting protein–ligand-binding sites experimentally is time-consuming and expensive. Thus, a variety of in silico methods to detect and predict binding sites was proposed as they can be scalable, fast and present low cost. Results We proposed Graph-based Residue neighborhood Strategy to Predict binding sites (GRaSP), a novel residue centric and scalable method to predict ligand-binding site residues. It is based on a supervised learning strategy that models the residue environment as a graph at the atomic level. Results show that GRaSP made compatible or superior predictions when compared with methods described in the literature. GRaSP outperformed six other residue-centric methods, including the one considered as state-of-the-art. Also, our method achieved better results than the method from CAMEO independent assessment. GRaSP ranked second when compared with five state-of-the-art pocket-centric methods, which we consider a significant result, as it was not devised to predict pockets. Finally, our method proved scalable as it took 10–20 s on average to predict the binding site for a protein complex whereas the state-of-the-art residue-centric method takes 2–5 h on average. Availability and implementation The source code and datasets are available at https://github.com/charles-abreu/GRaSP. Supplementary information Supplementary data are available at Bioinformatics online.

A perda de sincronismo é caracterizada através da ocorrência de eventos que provocam variações do ângulo de carga e aceleração ou frenagem do rotor e, dependendo da severidade da ocorrência, pode o sistema se tornar instável, não sendo possível o gerador retornar à velocidade síncrona. Portanto, dada a ocorrência deste evento, a máquina estará sujeita a diversas consequências que são capazes de ocasionar danos ao gerador síncrono e, consequentemente, prejuízos econômicos. Sendo assim, este trabalho aborda a implementação de um método inteligente visando a classificação da perda de sincronismo em geradores síncronos. Busca-se também realizar a comparação entre a técnica de classificação proposta e os métodos convencionais existentes utilizados em relés de proteção.

This work proposes the use of Genetic Algorithms (GA) to identify the area of the breast from the background in thermographic breast images. The proposed method uses color information, a fitness function based on cardioids, and GA. This is the first work in the literature to propose a Region of Interest (ROI) extraction based on GA and cariods. ROI extraction can improve the accuracy of cancer detection and assist with the standardization of acquisition protocols. The method is able to successfully separate the breast region in 52 out of 58 images, while being fully automatic, and not requiring manual selection of seed points.

Enzyme active sites are important and conserved functional regions of proteins whose identification can be an invaluable step toward protein function prediction. Most of the existing methods for this task are based on active site similarity and present limitations including performing only exact matches on template residues, template size restraints, despite not being capable of finding inter-domain active sites. To fill this gap, we proposed GASS-WEB, a user-friendly web server that uses GASS (Genetic Active Site Search), a method based on an evolutionary algorithm to search for similar active sites in proteins. GASS-WEB can be used under two different scenarios: (i) given a protein of interest, to match a set of specific active site templates; or (ii) given an active site template, looking for it in a database of protein structures. The method has shown to be very effective on a range of experiments and was able to correctly identify >90% of the catalogued active sites from the Catalytic Site Atlas. It also managed to achieve a Matthew correlation coefficient of 0.63 using the Critical Assessment of protein Structure Prediction (CASP 10) dataset. In our analysis, GASS was ranking fourth among 18 methods. GASS-WEB is freely available at http://gass.unifei.edu.br/.

The analysis of ambient (home, office, outdoor) atmosphere in order to check the presence of dangerous gases is getting more and more important. Therefore, tiny sensors capable to distinguish the presence of specific pollutants is crucial. Herein, a resistive sensor based on a carbon modified tin oxide nanowires, able to classify different gases and estimate their concentration, is presented. The C-SnO2 nanostructures are grown by chemical vapor deposition and then used as a conductometric sensor under a temperature gradient. The device works at lower temperatures than pure SnO2, with a better response. Five outputs are collected and combined to form multidimensional data that are specific of each gas. Machine learning algorithms are applied to these multidimensional data in order to teach the system how to recognize different gases. The six tested gases (acetone, ammonia, CO, ethanol, hydrogen, and toluene) are perfectly classified by three models, demonstrating the goodness of the raw sensor response. The gas concentration can also be estimated, with an average error of 36% on the low concentration range 1–50 ppm, making the sensor suitable for detecting the exceedance of the danger thresholds.

Significant efforts have been made in last decades concerning intelligent systems, particularly in vehicular applications. Despite several solutions proposed for intelligent systems on the electrical vehicle platform, this work aims to demonstrate a novel approach for hardware and software interaction that allows the development of different modes of operation. Some capabilities include cooperative mode to fully autonomous mode and vice versa. Furthermore, this proposed embedded system provides the basis for higher module integration from different levels of application. The proposed work reports real experiments carried out in a local urban environment using a medium electric vehicle platform which includes an embedded system with some embedded sensors to illustrate the validity of such platform in the context of the Cybernetic Transportation System (CTS).

Active sites are regions in the enzyme surface designed to interact with other molecules. Given their importance to enzyme function, active site amino acids are more conserved during evolution than the whole sequence, and can be a useful source of information for function prediction. For this reason, great effort has been put into identifying active sites in proteins. The majority of methods for this purpose uses an active site template of a protein of known function to search for similar structures into proteins of unknown function. In this direction, we recently proposed GASS (Genetic Active Site Search), a method based on an evolutionary algorithm to search for active sites in proteins. Although the method obtained very accurate results, its main strength and weakness are related to using only the spatial distance from the template to the protein to evaluate candidate sites. In this direction, this paper proposes MeGASS, a multi-objective version of GASS that also considers the depth of the residues when looking for active sites. This is important, as active sites are known for being closer to the protein surface to allow interactions with ligands. Results showed the depth attribute improves over the results of GASS, and its role into the method is worth further investigation.

Mais de 14 mil famílias de proteínas estão anotadas no Pfam (Protein Families Database), das quais cerca de 3.500 ainda têm suas funções desconhecidas. Testes experimentais são caros e demorados e, na sua ausência, estudos têm demonstrado que a função de uma proteína pode ser inferida com sucesso baseando-se similaridade da sequência ou da estrutura de uma proteína hipotética e proteínas de função conhecida. Uma maneira de predizer a função de uma proteína é através da busca dos sítios de ligação (binding sites). Sítios de ligação são regiões na superfície de uma enzima especialmente modeladas para interagir com outras moléculas. Devido à sua importância para a função da enzima, os aminoácidos do sítio ativo são mais conservados durante a evolução do que a sequência como um todo. Consequentemente, eles podem ser uma rica fonte de informações para a predição de função.Diversos métodos já foram propostos para identicar sítios ativos com base em templates. Porém, eles apresentam algumas limitações. Grande parte desses métodos não é capaz de lidar com mutações conservativas, onde enzimas com a mesma função podem variar em termos da composição dos aminoácidos do sítio ativo. Além disso, muitos deles não são capazes de identicar a cadeia ao qual um resíduo pertence ou restrigem a busca em termos de número de resíduos no template ou distâncias máximas entre o template e o sítio candidato.O principal objetivo desta tese é propor um novo método para a busca de sítios ativos basedos em templates utilizando algoritmos genéticos com base em dados estruturais. Para isso foi proposto o Genetic Active Site Search (GASS), um algoritmo genético modelado para utilizar informações estruturais de um sítio ativo template na busca de enzimas com sítios ativos similares. O método pode encontrar sítios ativos com resíduos em cadeias diferentes e é capaz de lidar com mutações conservativas, além de não impor quaisquer restrições quanto ao número de resíduos no sítio ativo e a distância entre eles. Os resultados do GASS foram comparados com os sítios catalíticos anotados no Catalytic Site Atlas (CSA) utilizando quatro diferentes conjuntos de dados. Quando comparado com outros métodos de busca de sítios catalíticos, os resultados mostraram que o GASS pôde identicar corretamente mais de 90% dos sítios pesquisados. Experimentos também foram realizados utilizando os dados de sítios de ligação dacompetição CASP 10 e, quando comparado com os 17 métodos participantes, o GASS apareceu em quarto lugar, embora não tenha sido inicialmente desenvolvido com este propósito. Currently, 25% of proteins annotated in the Protein Families Database (Pfam) have their function unknown. Experimental tests are expensive and time-consuming, and research has shown that the function of a protein can be successfully inferred based on the sequence or structure similarity of a hypothetical function and other functions of known function.A way of predicting the function of a protein is to consider its binding sites. Binding sites are regions in the surface of an enzyme designed to interact with other molecules. Due to its importance to enzyme function, the residues in the active site are more conserved than the sequence as a whole, providing important information for function prediction. Hence, active sites are a rich source of information for protein function prediction.Many methods have been previously proposed to identify active sites based on similarity. However, they do present some limitations, such as not being capable of dealing with conservative mutations (which occur when enzymes with the same function dier in terms of active site residues composition), having diculties in assigning the active siteto a chain or restricting the number of residues in the template. The main goal of this thesis is to propose a new method for searching for activesites similar using genetic algorithms based on protein structural data, namely Genetic Active Site Search (GASS). The method is based on a genetic algorithm, modeled to use structural information from an active site template in the search for enzymes with similar active sites. The method can nd active sites with residues in dierent chains and is ableto handle conservative mutations, apart from not imposing any restrictions on the number of residues in the active site and the distance between them. GASS results were compared with catalytic sites noted in the Catalytic Site Atlas (CSA) using four dierent data sets. When compared to other search methods of catalytic sites, the results showed that GASS identied correctly over 90% of the surveyed sites. Experiments were also performed using data of binding sites from the competitionCASP 10, and when compared with the 17 participants methods, GASS appeared in fourth, regardless of not being initially developed with this purpose.

Motivation: Currently, 25% of proteins annotated in Pfam have their function unknown. One way of predicting proteins function is by looking at their active site, which has two main parts: the catalytic site and the substrate binding site. The active site is more conserved than the other residues of the protein and can be a rich source of information for protein function prediction. This article presents a new heuristic method, named genetic active site search (GASS), which searches for given active site 3D templates in unknown proteins. The method can perform non-exact amino acid matches (conservative mutations), is able to find amino acids in different chains and does not impose any restrictions on the active site size. Results: GASS results were compared with those catalogued in the catalytic site atlas (CSA) in four different datasets and compared with two other methods: amino acid pattern search for substructures and motif and catalytic site identification. The results show GASS can correctly identify >90% of the templates searched. Experiments were also run using data from the substrate binding sites prediction competition CASP 10, and GASS is ranked fourth among the 18 methods considered. Availability and implementation: Source code and datasets (dcc.ufmg.br/ ∼glpappa/gass). Contact: sandroizidoro@unifei.edu.br Supplementary information: Supplementary data are available at Bioinformatics online.

A área de Aprendizado de Máquina Automático tem como objetivo recomendar automaticamente fluxos de tarefas que devem ser seguidas para criar algoritmos de aprendizado personalizados para uma dada base de dados. Essas tarefas incluem métodos de pré-processamentodedados, algoritmosdeaprendizadoeseusparâmetrosetécnicasde pós-processamento. Agrandevantagemdosmétodosdessaáreaestáemsuacapacidade de gerar fluxos sem dependência de conhecimento especializado do usuário realizando a tarefa. Esta dissertação propõe o RECIPE (REsilient ClassifIcation Pipeline Evolution), um método que faz uso de programação genética baseada em gramática para buscar por esses fluxos de tarefa considerando problemas de classificação. O RECIPE é flexível o suficiente para receber diferentes gramáticas e pode ser facilmente estendido para outras tarefas de aprendizado. Os resultados da medida F1 obtidos pelo RECIPE em 10 bases de dados são comparados a dois métodos estado da arte nessa tarefa, e são tão bons ou melhores do que os relatados anteriormente na literatura. Automatic Machine Learning is a growing area of machine learning that has a similar objective to the area of hyper-heuristics: to automatically recommend optimized pipelines, algorithms or appropriate parameters to specific tasks without much dependency on user knowledge. The background knowledge required to solve the task at hand is actually embedded into a search mechanism that builds personalized solutions to the task. Following this idea, this thesis proposes RECIPE (REsilient ClassifIcation Pipeline Evolution), a framework based on grammar-based genetic programming that builds customized classification pipelines. The framework is flexible enough to receive different grammars and can be easily extended to other machine learning tasks. RECIPE overcomes the drawbacks of previous evolutionary-based frameworks, such as generating invalid individuals, and organizes a high number of possible suitable data pre-processing and classification methods into a grammar. Results of f-measure obtained by RECIPE are compared to those two state-of-the-art methods, and shown to be as good as or better than those previously reported in the literature. RECIPE represents a first step towards a complete framework for dealing with different machine learning tasks with the minimum required human intervention.