Effective Feature Selection Method for Class-Imbalance Datasets Applied to Chemical Toxicity Prediction.

During the drug development process, it is common to carry out toxicity tests and adverse effect studies, which are essential to guarantee patient safety and the success of the research. The use of in silico quantitative structure-activity relationship (QSAR) approaches for this task involves processing a huge amount of data that, in many cases, have an imbalanced distribution of active and inactive samples. This is usually termed the class-imbalance problem and may have a significant negative effect on the performance of the learned models. The performance of feature selection (FS) for QSAR models is usually damaged by the class-imbalance nature of the involved datasets. This paper proposes the use of an FS method focused on dealing with the class-imbalance problems. The method is based on the use of FS ensembles constructed by boosting and using two well-known FS methods, fast clustering-based FS and the fast correlation-based filter. The experimental results demonstrate the efficiency of the proposal in terms of the classification performance compared to standard methods. The proposal can be extended to other FS methods and applied to other problems in cheminformatics.

Maximum common property: a new approach for molecular similarity.

The maximum common property similarity (MCPhd) method is presented using descriptors as a new approach to determine the similarity between two chemical compounds or molecular graphs. This method uses the concept of maximum common property arising from the concept of maximum common substructure and is based on the electrotopographic state index for atoms. A new algorithm to quantify the similarity values of chemical structures based on the presented maximum common property concept is also developed in this paper. To verify the validity of this approach, the similarity of a sample of compounds with antimalarial activity is calculated and compared with the results obtained by four different similarity methods: the small molecule subgraph detector (SMSD), molecular fingerprint based (OBabel_FP2), ISIDA descriptors and shape-feature similarity (SHAFTS). The results obtained by the MCPhd method differ significantly from those obtained by the compared methods, improving the quantification of the similarity. A major advantage of the proposed method is that it helps to understand the analogy or proximity between physicochemical properties of the molecular fragments or subgraphs compared with the biological response or biological activity. In this new approach, more than one property can be potentially used. The method can be considered a hybrid procedure because it combines descriptor and the fragment approaches.

In silico study of the human rhodopsin and meta rhodopsin II/S-arrestin complexes: impact of single point mutations related to retina degenerative diseases.

We propose two models of the human S-arrestin/rhodopsin complex in the inactive dark adapted rhodopsin and meta rhodopsin II form, obtained by homology modeling and knowledge based docking. First, a homology model for the human S-arrestin was built and validated by molecular dynamics, showing an average root mean square deviation difference from the pattern behavior of 0.76 A. Then, combining the human S-arrestin model and the modeled structure of the two human rhodopsin forms, we propose two models of interaction for the human S-arrestin/rhodopsin complex. The models involve two S-arrestin regions related to the N domain (residues 68-78; 170-182) and a third constituent of the C domain (248-253), with the rhodopsin C terminus (330-348). Of the 22 single point mutations related to retinitis pigmentosa and congenital night blindness located in the cytoplasmatic portion of rhodopsin or in S-arrestin, our models locate 16 in the interaction region and relate two others to possible dimer formation. Our calculations also predict that the light activated complex is more stable than the dark adapted rhodopsin and, therefore, of higher affinity to S-arrestin.

Quantitative study of the structure-retention index relationship in the imine family.

The Kováts retention index is one of the most popular descriptors of the performance of organic compounds in gas chromatography (GC). The mathematical modeling of this index is an interesting and open problem in analytical chemistry. In this paper, two models for the prediction of the Kováts retention index are presented. Topologic, topographic and quantum-chemical descriptors were used as structural descriptors. Multiple linear regression (MLR) analysis provides the first model using the forward stepwise procedure for the variable selection. For the second one, an ensemble of artificial neural network (ANN) was constructed using the pruning algorithm. Both methods were validated by an external set of compounds, by the Golbraikh and Tropsha method and by the leave-one-out (LOO) and the leave many out (LMO) procedures. The R2, RMScv and Q2, values for the training sets were 0.884, 0.589 and 0.830 for NN and 0.974, 0.417 and 0.970 for MLR models, respectively. The robustness of both models was demonstrated. Both portrait the chromatographic performance of the sample but in this case, the results of MLR equation are better than the NN ones. The MLR model is recommended because of its simplicity.

Definition of a novel atomic index for QSAR: the refractotopological state.

PURPOSE: In this work, a novel descriptor of atoms in molecules is introduced. The Refractotopological State Index for atoms (R-state,R), rectifies the atomic refractivity values reported by Crippen et al with the atomic refractivity values of the topological environment of each skeletal atom in the molecule. METHOD: The R-state (R(i)), for atom i is an hybrid index that is defined as the intrinsic refractivity value of the atom i (AR(i)) plus a perturbation term Delta AR(i) in the non-hydrogen depleted graph. RESULTS: The variations of the R values in different molecules are showed. QSAR examples previously reported by other authors are given for benzimidazole inhibition of Lee strain flu virus and receptor binding affinity of beta-carbolines. CONCLUSIONS: The index does not only describe the representation of the atomic dispersive forces related to the molar refractivity but also the influence of bounded and unbounded atoms as a measure of the distance-effect of the other groups in the molecule. The R-state index has proved a good performance, either alone or combined with the electro topological (E)-state index. This implies that in those cases this representation of dispersive forces between the molecule and the active site is a valid approach to the biological problem

Similitud molecular empleando Índices Híbridos / Molecular similarity using Hybrid Indices

Se presenta un método para la detección de semejanza entre moléculas basado en macheo inexacto de grafos. Se parte del grafo molecular completo ponderado en sus vértices por propiedades químico-físicas particionadas sobre los mismos, se reduce el grafo por el procedimiento CALEDE que define Centros Descriptores o fragmentos de primer orden, los cuales son subgrafos ponderados por la suma de los valores de los vértices ponderados individualmente a su vez, y se construyen fragmentos denominados de segundo orden que incluyen la distancia entre los centros de masas de ambos centros descriptores. Se presenta el método de búsqueda aplicado a una base de datos de más de 300 moléculas con sus respectivas estructuras en tres dimensiones. Esos compuestos se encuentran evaluados como anticancerígenos en la base de datos de compuestos del NCBI-USA. En el experimento computacional se encuentra que, en dependencia de la función de similitud empleada, es posible detectar compuestos que a pesar de poseer diferente topología, poseen valores de las propiedades empleadas para el macheo lo cual sugiere la presencia de potenciales farmacóforos como hallazgo relevante, lo cual constituiría un novedoso enfoque para el diseño computacional de fármacos(AU)

A method for detecting similarity between molecules based on inexact matching graph is presented. We start from a complete molecular graph vertices weighted by several hybrid indices. The molecular graph is reduced by CALEDE procedure, which define descriptors centers or first order fragments. These fragments are subgraphs weighted with the sum of values of the vertices weighted with the hybrid indices. It also define second order fragments by including the distance between the centers of mass of both descriptors centers. The search method applied to a database of over 300 molecules with their respective threedimensional structures is presented. These compounds are reported in the NCBI-USA database of compounds whish were evaluated in anticancer tests. In the computational experiment, depending on the similarity function used, is possible to detect compounds that despite having different topology have property values suggesting the presence of potential pharmacophore. It suggest the possibility to use this approach as a novel approach for computational drug design(AU)

Real-time external labeling layout algorithm for Direct Volume Rendering

Illustrations used in technical and scientific texts often employ labels to correlate the graphic elements and their textual descriptions. Researchers have proposed several algorithms to determine the layout of the annotations on images rendered at interactive frame rates. Generally these layouts can be classified as internal or external. This paper proposes a new algorithm for locating external labels during the real-time direct rendering of volume data. The proposed algorithm uses only the rows of pixels corresponding to the labels anchor points, which optimizes the performance and facilitates its implementation, avoiding the computation of the convex hull for the generated image. Both, the overall visualization performance and the cost of the proposed algorithm are kept in real-time (60 fps) for medium size volumes (about 2563 voxels)(AU).

Las ilustraciones utilizadas en documentos científicos y técnicos utilizan frecuentemente etiquetas para correlacionar los elementos gráficos y sus textos descriptivos. Los investigadores han propuesto diversos algoritmos para determinar el posicionamiento en tiempo real de las correspondientes anotaciones en las imágenes obtenidas en un marco interactivo. Generalmente estos posicionamientos se clasifican como internos o externos. Este artículo propone un nuevo algoritmo para ubicar etiquetas externas en tiempo real durante la obtención de datos de volumen. El algoritmo propuesto usa solo las filas de píxels correspondientes a los puntos de presentación de las etiquetas lo que optimiza el desempeño y facilita la implementación haciendo innecesarios algunos cálculos. Tanto el desempeño general de la vista como el costo del algoritmo propuesto se obtienen en tiempo real (60 fps) para volúmenes de mediana talla (alrededor de 256 voxels)(AU)

Sistema de visualización remota para la representación interactiva de volúmenes de datos médicos / Remote visualization system for interactive representation of medical volume dataset

Las aplicaciones de visualización médica han adquirido un elevado auge en la medicina a nivel mundial, ya que les permite a los médicos especialistas realizar diagnósticos preoperatorios no invasivos y de alta precisión desde una perspectiva 3D. La idea principal de la misma es obtener un modelo tridimensional de alta resolución gráfica a partir de imágenes médicas digitales de las modalidades de Tomografía Axial Computarizada y Resonancia Magnética Nuclear. Los usuarios de este tipo de aplicaciones demandan de forma creciente que las aplicaciones permitan el diagnóstico de patologías en un entorno de trabajo colaborativo. En este trabajo presentamos una arquitectura para sistemas de visualización remota basados en la transmisión de imágenes. El esquema de comunicación y transmisión de datos e imágenes entre el servidor y los clientes utiliza RTP como protocolo de comunicación. Los resultados obtenidos demuestran que la variante de algoritmo RLE implementada permite obtener visualizaciones interactivas y en tiempo real con un consumo mínimo del ancho de banda de la red(AU)

Three-dimensional medical visualization applications have acquired a high rise in medicine. They allow specialized doctors to make preoperative diagnostics with high accuracy from a 3D perspective. The main idea of medical visualization is to obtain a three-dimensional and high-resolution graphics from digital medical imaging modalities like computed tomography and magnetic resonance imaging. The users of these applications increasingly demand that applications allow diagnosis in a collaborative work environment. Architecture for remote visualization systems based on image is presented. The server and client scheme of communication and transmission of data and images use RTP as communication protocol. Our results show that the implemented variant of RLE algorithm allows interactive and real time representation with a minimum of bandwidth(AU)

Funciones de transferencia en el perceptrón multicapa: efecto de su combinación en entrenamiento local y distribuido / Transfer functions in the multilayer perceptron: effects of its combination on local and distributed training

El perceptrón multicapa (PMC) figura dentro de los tipos de redes neuronales artificiales (RNA) con resultados útiles en los estudios de relación estructura-actividad. Dado que los volúmenes de datos en proyectos de Bioinformática son eventualmente grandes, se propuso evaluar algoritmos para acortar el tiempo de entrenamiento de la red sin afectar su eficiencia. Se desarrolló un algoritmo para el entrenamiento local y distribuido del PMC con la posibilidad de variar las funciones de transferencias para lo cual se utilizaron el Weka y la Plataforma de Tareas Distribuidas Tarenal para distribuir el entrenamiento del perceptrón multicapa. Se demostró que en dependencia de la muestra de entrenamiento, la variación de las funciones de transferencia pueden reportar resultados mucho más eficientes que los obtenidos con la clásica función Sigmoidal, con incremento de la g-media entre el 4.5 y el 17 por ciento. Se encontró además que en los entrenamientos distribuidos es posible alcanzar eventualmente mejores resultados que los logrados en ambiente local(AU)

The multilayer perceptron (PMC) ranks among the types of artificial neural networks (ANN), which has provided better results in studies of structure-activity relationship. As the data volumes in Bioinformatics' projects are eventually big, it was proposed to evaluate algorithms to shorten the training time of the network without affecting its efficiency. There were evaluated different tools that work with ANN and were selected Weka algorithm for extracting the network and the Platform for Distributed Task Tarenal to distribute the training of multilayer perceptron. Finally, it was developed a training algorithm for local and distributed the MLP with the possibility of varying transfer functions. It was shown that depending on the training sample, the change of transfer functions can yield results much more efficient than those obtained with the classic sigmoid function with increased g-media between 4.5 and 17 percent. Moreover, it was found that with distributed training can be achieved eventually, better results than those achieved in the local environment(AU)