Single Descriptor Research Articles

Application of two methods of calculation of solvation descriptor L to estimate C5 -C7 alkenes retention.

The solvation descriptor L for 59 isomers of all C5 -C7 alkenes was calculated using two methods based on additive contributions of particular fragments in the molecule by the method of Havelec and Ševčík and the method of Platts and Butina. These descriptors were used to estimate the gas chromatography retention of alkenes on squalane and polydimethylsiloxane stationary phases. The retention was described better by the Platts-Butina method. Modification of the Havelec-Ševčík method by omitting the contribution for interaction of the cis isomers led to a substantial improvement in the estimation ability of the model. The modified Havelec-Ševčík method was found to be preferable for estimation of the descriptor L compared to the Platts-Butina method. A more comprehensive description of the retention of alkenes was achieved by inclusion of an additional descriptor E. This model with the descriptors L and E yielded better estimation for alkenes compared to the model with a single descriptor.

Simplified molecular input line entry system-based: QSAR modelling for MAP kinase-interacting protein kinase (MNK1)

Quantitative structure–activity relationship (QSAR) models were built for the prediction of inhibition (pIC50, i.e. negative logarithm of the 50% effective concentration) of MAP kinase-interacting protein kinase (MNK1) by 43 potent inhibitors. The pIC50 values were modelled with five random splits, with the representations of the molecular structures by simplified molecular input line entry system (SMILES). QSAR model building was performed by the Monte Carlo optimisation using three methods: classic scheme; balance of correlations; and balance correlation with ideal slopes. The robustness of these models were checked by parameters as rm2, r*m2, and randomisation technique. The best QSAR model based on single optimal descriptors was applied to study in vitro structure–activity relationships of 6-(4-(2-(piperidin-1-yl) ethoxy) phenyl)-3-(pyridin-4-yl) pyrazolo [1,5-a] pyrimidine derivatives as a screening tool for the development of novel potent MNK1 inhibitors. The effects of alkyl group, –OH, –NO2, F, Cl, Br, I, etc. on the IC50 values towards the inhibition of MNK1 were also reported.

(Invited) Water As Promoter and Catalyst in Dioxygen Electrochemistry at Aqueous and Organic Electrified Interfaces

Water and di-oxygen are the most important and pervasive chemicals that influence all areas of science, from physics to biology to chemistry, and are particularly important in governing phenomena occurring at solid-liquid interfaces in electrochemical environments. One property that sets aqueous-based electrochemical interfaces apart from all other interfaces is the unique arrangements of water molecules within the distinctive interactions between metal surfaces and hydrated ions. These interactions are complex, involving molecular and dissociative adsorption, as well as van der Waals forces both between water molecules and neighboring ions in the compact part of the double layer (i.e. solvation) and between water and covalently-bonded co-adsorbates (e.g. adsorbed hydroxyl species, OHad). Despite the enormous fundamental and technological significance in unfolding these complexities, previous work has treated the water molecule itself simply as a reactant needed to satisfy the stoichiometry of electrochemical reactions, rather than a vital hydrogen-/hydroxyl-donor molecule that can promote the rates of transformation of reaction intermediates to final products. The prime example of such an “incomplete” treatment of water is the oxygen reduction reaction (ORR) in “proton”-O2 fuel cell systems and in lithium-O2 battery systems. Current models guiding the development of new materials for these systems rely on a single descriptor - the binding energy of oxygen intermediates – that cannot explain many “anomalous” phenomena. Here, we introduce a new model in which water itself plays a critical role in determining the reaction kinetics in dioxygen electrochemistry. The model is based on the formation of HOad···H2O (aqueous solutions), and LiO2···H2O (organic solvents) complexes that place water in a configurationally favorable position for proton transfer to weakly adsorbed intermediates. We found that water plays a dual role, acting as a key promoter in both environments as well as a catalyst in Li-O2 systems

Nonlinear QSAR models with high-dimensional descriptor selection and SVR improve toxicity prediction and evaluation of phenols on Photobacterium phosphoreum

Assessment of the risk of chemicals is an important task in the environmental protection. In this paper, we developed quantitative structure–activity relationship (QSAR) methods to evaluate the toxicity of phenol to Photobacterium phosphoreum, which is an important indicator for water quality. We first built support vector regression (SVR) model using three descriptors, and the SVR model (t=2) had the highest external prediction ability (MSEext=0.068, Qext2=0.682), about 40% higher than literature model's. Second, to identify more effective descriptors, we applied in-house methods to select descriptors with clear meanings from 2835 descriptors calculated by the PCLIENT and used them to construct the SVR models. Our results showed that our twenty new QSAR models significantly increased the standard regression coefficient on test set (MSEext values ranged from 0.003 to 0.063 and Qext2 values ranged from 0.708 to 0.985). The Y random response permutation test and different splits of training/test datasets also supported the excellent predictive power of the best SVR model. We further evaluated the regression significance of our SVR model and the importance of each single descriptor of the model according to the interpretability analysis. Our work provided useful theoretical understanding of the toxicity of phenol analogues.

An integrated approach for image retrieval using local binary pattern

In this paper an integrated approach for image retrieval has been proposed that uses the concept of local binary pattern. The image is divided into a fixed number of blocks and from each block, LBP based color, texture and shape features are computed. LBP histogram is used for the extraction of color and texture features. Region code based scheme is used to support region based retrieval. Center pixel and its neighbors are used to improve the discrimination power of Local Binary Patterns. Shape feature computed using the binary edge map obtained using Sobel edge detector is combined with color and texture features to make a single completed binary region descriptor. To support region based retrieval, a more effective region code based scheme is employed. The approach is tested on different benchmark databases like COREL, CIFAR-10 and MPEG-7 CCD database. The experimental results have verified that the proposed scheme has impressive retrieval performance in comparison to state-of-the-art techniques.

Uniting Keypoints: Local Visual Information Fusion for Large-Scale Image Search

In this paper, we propose a novel approach to address the problem of the huge amount of local features for a large-scale database. First, in each image the local features are organized into dozens of groups by performing the standard $k$ -means clustering algorithm on their spatial positions. Second, a compact descriptor is generated to describe the visual information of each group of local features. Since, in each image, thousands of local features are reorganized into only dozens of groups and each group is described by a single descriptor, the total amount of descriptors in a large-scale database will be greatly reduced. Therefore, we can reduce the complexity of the searching procedure significantly. Further, the generated group descriptors are encoded into binary format to achieve the storage and computation efficiency. The experiments on two benchmark datasets, i.e., UKBench and Holidays, with the Flickr1M distractor database demonstrate the effectiveness of the proposed approach.

Cluj and Omega Polynomials in PAHs and Fullerenes

A graph property P(G), written as a sequence of numbers counting its local partitions, can be expressed as a counting polynomial P(G,x) with the exponents of the indeterminate x showing the extent of partitions p(G), υ p(G) = P(G) while the coefficients are related to the number of partitions occurrence. Definitions and properties of Cluj CJ(G, x) and Omega Ω(G, x) polynomials and their relation with other topological descriptors are presented. Analytical relations for calculating these polynomials and corresponding topological indices in some classes of PAHs are derived. The ability of topological descriptors to predict various physico-chemical and biological properties is reviewed. Omega polynomial applied to small fullerenes provided a single number descriptor useful in predicting their energetic stability.

On-the-fly learning for visual search of large-scale image and video datasets.

The objective of this work is to visually search large-scale video datasets for semantic entities specified by a text query. The paradigm we explore is constructing visual models for such semantic entities on-the-fly, i.e. at run time, by using an image search engine to source visual training data for the text query. The approach combines fast and accurate learning and retrieval, and enables videos to be returned within seconds of specifying a query. We describe three classes of queries, each with its associated visual search method: object instances (using a bag of visual words approach for matching); object categories (using a discriminative classifier for ranking key frames); and faces (using a discriminative classifier for ranking face tracks). We discuss the features suitable for each class of query, for example Fisher vectors or features derived from convolutional neural networks (CNNs), and how these choices impact on the trade-off between three important performance measures for a real-time system of this kind, namely: (1) accuracy, (2) memory footprint, and (3) speed. We also discuss and compare a number of important implementation issues, such as how to remove ‘outliers’ in the downloaded images efficiently, and how to best obtain a single descriptor for a face track. We also sketch the architecture of the real-time on-the-fly system. Quantitative results are given on a number of large-scale image and video benchmarks (e.g. TRECVID INS, MIRFLICKR-1M), and we further demonstrate the performance and real-world applicability of our methods over a dataset sourced from 10,000 h of unedited footage from BBC News, comprising 5M+ key frames.

A supervised dictionary learning and discriminative weighting model for action recognition

In this paper, we propose a supervised dictionary learning algorithm for action recognition in still images followed by a discriminative weighting model. The dictionary is learned based on Local Fisher Discrimination which takes into account the local manifold structure and discrimination information of local descriptors. The label information of local descriptors is considered in both dictionary learning and sparse coding stage which generates a supervised sparse coding algorithm and makes the coding coefficients discriminative. Instead of using spatial pyramid features, sliding window-based features with max-pooling are computed from coding coefficients. And then a discriminative weighting model combining a max-margin classifier is proposed using the features. Both the weighting coefficients and model parameters can be jointly learned using the same way in Multiple Kernel Learning algorithm. We validate our model on the following action recognition datasets: Willow 7 human actions dataset, People Playing Music Instrument (PPMI) dataset, and Sports dataset. To show the generality of our model, we also validate it on Scene15 dataset. The experiment results show that only with single scale local descriptors, our algorithm is comparable to some state-of-the-art algorithms.

Computer vision for virus image classification

In this paper we present a new ensemble of descriptors for the classification of transmission electron microscopy images of viruses that is based on texture analysis. A set of six well-assessed texture descriptors, namely Local Binary Pattern ( LBP ), Local Ternary Pattern ( LTP ), Dense LBP ( DLBP ), Rotation Invariant co-occurrence among LBP ( RI ), Local Phase Quantization ( LPQ ), and LBP Histogram Fourier ( LHF ), are combined with innovative approaches to improve their performances in virus classification. To generate new variants of the aforementioned descriptors, different approaches are applied: (i) the Edge approach ( ED ), which extracts the textural information from specific regions of the image instead of from the original image; (ii) the Bag of Features ( BOF ) scheme, which is used to build vocabularies of the most representative patterns; and (iii) Multi-Quinary coding ( MQ ). Moreover, to demonstrate the generalizability and applicability of ED and BOF , they are tested on additional datasets containing subcellular parts and tissues. We observe that ED improves the performance of the single standard descriptors, while BOF is particularly effective in fusion by sum rule with the standard application of texture descriptors (i.e., when they are extracted from the entire image). In contrast, the combination of MQ with LHF , DLBP , and RI does not improve virus classification. Using these results, we suggest a new ensemble of descriptors called NewF , which is based on the best methods investigated in this paper, as well as on some other state-of-the-art descriptors. NewF accuracy in virus classification is 85.7%, outperforming previous methods proposed in the literature for the same task (i.e., virus classification using the object scale dataset). The MATLAB code for our methods and NewF descriptor are available at https://www.dei.unipd.it/node/2357 .

Efficient multi-modal fusion on supergraph for scalable image annotation

Different types of visual features provide multi-modal representation for images in the annotation task. Conventional graph-based image annotation methods integrate various features into a single descriptor and consider one node for each descriptor on the learning graph. However, this graph does not capture the information of individual features, making it unsuitable for propagating the labels of annotated images. In this paper, we address this issue by proposing an approach for fusing the visual features such that a specific subgraph is constructed for each visual modality and then subgraphs are connected to form a supergraph. As the size of supergraph grows linearly with the number of visual features, it is essential to handle large computational complexity of label propagation on the supergraph. To this end, we extract some prototypes from the feature vectors of images and incorporate them into the supergraph construction. The learning process is then conducted on the prototypes, instead of a large number of feature vectors, making the label inference scalable. The experiments on a wide range of standard datasets reveal that the proposed approach achieves scalable image annotation while having an acceptable level of performance.

A single theoretical descriptor for the bond-dissociation energy of substituted phenols

Relative to the corresponding value of phenol, the bond-dissociation energies (BDE) of substituted phenols correlate well with a single descriptor: the Mulliken charge on the oxygen atom of the phenoxyl radical. However, the correlation fails for phenols ortho-substituted with polar groups. Internal reaction coordinates (IRC) for the model reaction of hydrogen abstraction by the hydroperoxyl radical from various 2- and 4-substituted phenols were calculated in order to investigate the role of intra-molecular hydrogen bonds and steric effects on the process. Calculations yielded theoretical values in good agreement with experimental ΔBDE values. The hydrogen-abstraction process was further analyzed in terms of density functional theory (DFT)-based reactivity indices such as local electrophilicity, the Fukui function for nucleophilic attack, and dual descriptor values of the phenolic hydroxyl oxygen along the IRC.

Fuzzy decision tree to classify complex fractionated atrial electrograms.

Catheter ablation has emerged as an effective treatment strategy for atrial fibrillation (AF) in recent years. During AF, complex fractionated atrial electrograms (CFAE) can be recorded and are known to be a potential target for ablation. Automatic algorithms have been developed to simplify CFAE detection, but they are often based on a single descriptor or a set of descriptors in combination with sharp decision classifiers. However, these methods do not reflect the progressive transition between CFAE classes. The aim of this study was to develop an automatic classification algorithm, which combines the information of a complete set of descriptors and allows for progressive and transparent decisions. We designed a method to automatically analyze CFAE based on a set of descriptors representing various aspects, such as shape, amplitude and temporal characteristics. A fuzzy decision tree (FDT) was trained and evaluated on 429 predefined electrograms. CFAE were classified into four subgroups with a correct rate of 81±3%. Electrograms with continuous activity were detected with a correct rate of 100%. In addition, a percentage of certainty is given for each electrogram to enable a comprehensive and transparent decision. The proposed FDT is able to classify CFAE with respect to their progressive transition and may allow objective and reproducible CFAE interpretation for clinical use.

Application of Machine Learning Method in Genomics and Proteomics

With the avalanche of genomic and proteomic data generated in the postgenomic age, it is highly desirable to develop automated methods for rapidly and effectively analyzing and predicting the structure, function, and other properties of DNA and protein. The machine learning methods have become an important strategy for the discovery of potential knowledge in genomics and proteomics. Researches in recent years have shown tremendous advances in the properties prediction of DNA fragments and protein sequences by various pattern recognition methods. These techniques provide economical and timesaving solutions for identifying the properties of DNA and protein. This special issue was hosted for the recent development of the application of machine learning methods in genomics and proteomics. In this special issue, five works focused on the protein classification. How to extract key features from a protein was a key step in the discrimination of protein class. B. Liu et al. proposed to use Position-Specific Score Matrix (PSSM) and Accessible Surface Area (ASA) to formulate protein samples. The hidden Markov support vector machine (HM-SVM) was employed to predict protein binding site. Simulation in fivefold cross-validation on a benchmark dataset including 1124 protein chains showed that their method is more accurate for protein binding site prediction than some state-of-the art methods. This method can also be applied in DNA binding site, vitamin binding site, and posttranslational modification of proteins. Based on chemical shift (CS) information derived from nuclear magnetic resonance (NMR), F. Yonge proposed a novel feature to predict protein supersecondary structures. The quadratic discriminant (QD) analysis was selected as the prediction algorithm. Overall accuracy in threefold cross-validation is 77.3% for predicting four types of supersecondary structures. According to the concept of pseudo amino acids, G.-L. Fan et al. proposed the average chemical shifts (ACS) composition and established an online webserver called acACS which was calculated from average chemical shift information and protein secondary structure. By using SVM as the classification algorithm, the acACS was used in the discrimination between acidic and alkaline enzymes and between bioluminescent and nonbioluminescent proteins. Encouraging results were achieved. The protein secondary structure, structure class, and disorder region can be predicted using the AC-based method. L. Nanni et al. proposed to combine different features to improve protein prediction. These features include amino acids composition, PSSM, and substitution matrix representation (SMR). Each feature is used to train a separate SVM. Total of 15 benchmark datasets were used to evaluate the performance of their proposed method. Comparative results show that the PSSM always produces good accuracies. However, no single descriptor is superior to all others across all test datasets. The major contribution in this paper is to propose an ensemble of classifiers for sequence-based protein classification. H. Lin et al. briefly reviewed the development of ion channel prediction using machine learning method. They initially introduced how to construct a valid and objective benchmark dataset to train and test the predictor. Subsequently, the mathematical descriptors were presented to formulate the ion channel sequences. Moreover, two feature selection techniques on how to optimize feature set were described. Finally, the support vector machine was suggested performing classification. The methods introduced in that review can be generalized into other protein prediction fields as well. The paper from P. Feng et al. was the unique work focused on DNA prediction using machine learning method. They proposed a novel descriptor called pseudo K-tuple nucleotide composition (PseKNC) to formulate the DNA sequences. The feature is calculated from K-tuple nucleotide composition and the structural correlation of DNA dinucleotides. Subsequently, the SVM was used to predict DNase I hypersensitive sites. The jackknife cross-validated accuracy is 83%, which is competitive with that of the existing method. This new descriptor can also be widely used in DNA regulatory elements prediction. Hao Lin Wei Chen Ramu Anandakrishnan Dariusz Plewczynski

Quantitative relationships for the prediction of the vapor pressure of some hydrocarbons from the van der Waals molecular surface

A quantitative structure - property relationship (QSPR) modeling of vapor pressure at 298.15 K, expressed as log (VP / Pa) was performed for a series of 84 hydrocarbons (63 alkanes and 21 cycloalkanes) using the van der Waals (vdW) surface area, SW/Å2, calculated by the Monte Carlo method, as the molecular descriptor. The QSPR model developed from the subset of 63 alkanes (C1-C16), deemed as the training set, was successfully used for the prediction of the log (VP / Pa) values of the 21 cycloalkanes, which was the external prediction (test) subset. A QSPR model was also developed for a series composed of all 84 hydrocarbons. Both QSPR models were statistically tested for their ability to fit the data and for prediction. The results showed that the vdW molecular surface used as molecular descriptor (MD) explains the variance of the majority of the log (VP / Pa) values in this series of 84 hydrocarbons. This MD describes very well the intermolecular forces that hold neutral molecules together. The clear physical meaning of the molecular surface values, SW/Å2, could explain the success of the QSPR models obtained with a single structural molecular descriptor.

Magnetic resonance imaging of intraductal papillomas: typical findings and differential diagnosis.

Even upon core biopsy, accurate classification of benign intraductal papillomas (IPs) can be difficult. Accordingly, IPs are still frequently surgically resected. Therefore, accurate classification of IP by magnetic resonance mammography (MRM) would potentially optimize patient management. However, the few investigations assessing MRM of IP included small patient collectives, and overall accuracy is still unknown. We performed this investigation to analyze the morphologic and dynamic MRM profiles of IP in more detail and to identify the overall accuracy of MRM for differential diagnosis of IP versus malignant breast lesions. Consecutive patients scheduled for MRM (standardized scanning protocols: dynamic T1-weighted gradient echo before/after Gd-DTPA [gadolinium diethylenetriamine pentaacetate; 0.1 mmol/kg body weight]; T2-weighted turbo spin echo) with subsequent surgicopathologic verification were enrolled. For the detailed assessment of morphologic and dynamic profiles, 2 experienced radiologists (>500 MRM examinations; blinded to surgicopathologic verification) performed prospective evaluation of MRM, in consensus, applying 17 predefined MRM descriptors. From this database, all patients showing IP (n = 83) or malignant breast lesions (n = 648) were further evaluated statistically: univariate analyses (association of single descriptors with IP/breast cancer: contingency table statistics) and multivariate analyses were performed to identify accurate descriptor combinations (CHAID [CHi-squared Automatic Interaction Detection]) and overall accuracy of MRM for differential diagnosis of IP versus malignant breast lesions (logistic regression; receiver operating characteristics [ROC], area under the ROC curve). There were 82.4% of MRM descriptors significantly associated with IP (n = 14; P < 0.05). The accuracy of single descriptors (odds ratio [OR], ≤10.6) could be further increased by descriptor combinations (double combination: OR ≤12.7; triple combination: OR ≤15.0). With area under the ROC curve = 0.90, there was a high overall accuracy of MRM for the differential diagnosis of IP versus malignant breast lesions. A detailed assessment of MRM allows precise characterization of benign IPs and accurate differentiation from malignant breast lesions.

QSPR studies on refractive indices of structurally heterogeneous polymers

We developed a predictive Quantitative Structure–Property Relationship (QSPR) for the refractive indices of 234 structurally diverse polymers. The model involves a single molecular descriptor and a conformation-independent approach. The most appropriate polymer structure representation was investigated by considering 1–5 monomeric repeating units. The established equations were validated and tested through various well-known techniques, such as the use of an external test set of compounds, the Cross-Validation method, Y-Randomization and Applicability Domain, and finally a comparison was also performed to published results from the li terature. The developed QSPR could be useful for assisting the development of new polymeric materials.

A Hybrid Bags-of-Feature model for Sports Scene Classification

In this paper, we propose an effective framework for sports scene classification by using a Hybrid Bags-of-Feature model. The Bag-of-Feature (BoF) model is a methodology which represents an image based on the histogram of visual codewords. With its successful performance, the BoF model has been widely exploited in computer vision such as scene classification and object recognition. Unlike traditional BoF models which resort to a single feature descriptor such as SIFT and thus exploit a single codebook, the proposed Hybrid BoF model employs two different types of codebooks; SIFT feature and modified LBP feature. The LBP descriptor is capable of capturing micro patterns of the image, hence it is suitable for texture classification. In contrast, Gradient based feature descriptors such as SIFT have been proven to be effective for object recognition. More specifically, foreground objects (or regions containing distinct silhouettes) are encoded based on the SIFT descriptor and the remaining regions such as sky, lawn or floor of the gymnasium are encoded based on the LBP descriptor. To build a criterion for descriptor selection, we introduce a saliency pyramid based on the Phase Fourier Transform (PFT). The proposed method has been extensively tested, and experimental results show that the proposed framework is effective for sports scene classification compared to other various state-of-the-art methods.

A universal comparison study of chromatographic response functions

We report on a large scale in silico comparison study of so-called chromatographic response functions (CRFs). These are single number descriptors of the separation quality that can be used to guide search-based optimizations for chromatographic separations. A comprehensive set of literature and new CRFs were compared for their ability to guide a search based on first order chromatographic data (i.e., no spectral information available) and for cases where the number of sample compounds is not known beforehand. The results are discussed based on the available separation power. It was found that CRFs increasing monotonically with the number of observed peaks perform significantly better than those that do not possess this property. CRFs based on the discrimination factor or the peak-to-valley ratio can better cope with peak asymmetry than CRFs based on Snyder resolution Rs. Unfortunately, the former lose their advantage as soon as the noise level becomes significant. Most CRFs perform best when the search is conducted on a column offering just, or, even better, a bit less than the required efficiency to baseline separate the sample. The best results over the entire range of possible efficiencies are obtained with a CRF giving preference to the number of observed compounds before further ranking the conditions based on the achieved separation resolution or the required analysis time. When the search is conducted on columns with an insufficient efficiency, even the best possible CRFs suffer from the incomplete information about the sample, and deviating searches cannot be avoided without resorting to spectral information of the sample.

Acolhimento em saúde: uma revisão sistemática em periódicos brasileiros

Este estudo apresenta uma revisão sistemática sobre acolhimento em saúde em periódicos brasileiros. As bases de dados pesquisadas foram Bireme, MedLine, PubMed, SciELO e Science Direct. Utilizando como descritor único o termo "acolhimento", a pesquisa resultou na análise quantitativa e qualitativa de 40 artigos publicados entre 1999 e 2011. Os resultados indicaram que: os dois periódicos com maior número de publicações sobre o tema foram Cadernos de Saúde Pública e Ciência &amp; Saúde Coletiva (25,0 e 22,5%, respectivamente); nos anos de 2004 a 2010, foram publicados sete artigos sobre o tema, e, em 2011, seis artigos foram encontrados. As áreas predominantes foram Saúde Pública/Coletiva (57,6%) e Enfermagem (39,4%). A análise qualitativa da variável "objeto de estudo" identificou artigos cuja investigação era os pontos de vista de usuários, profissionais e de ambos. Já a variável "percepções dos profissionais" ressaltou aspectos positivos do acolhimento. São necessários novos estudos que qualifiquem o conhecimento a respeito dessa forma de atendimento à população tanto nos serviços de atenção básica quanto de atenção especializada.