Optimización basada en mallas variables para predecir el plegamiento de proteínas
Fecha
2011-07-04
Autores
Moré Quintero, Iosvani
Título de la revista
ISSN de la revista
Título del volumen
Editor
Universidad Central “Marta Abreu” de Las Villas
Resumen
El presente trabajo describe el plegamiento de proteínas que constituye uno de los retos más importantes dentro de la biología molecular. Debido a que este problema es extremadamente difícil aún no se conoce un método capaz de resolverlo eficientemente.
Por lo anterior este informe propone un algoritmo de Optimización basado en Mallas Variables (VMO), para tratar dicha problemática. El algoritmo surge en el grupo de Inteligencia Artificial de nuestra universidad como un modelo poblacional. El modelo VMO utiliza un proceso de expansión formado por funciones de generación de nuevas soluciones que exploran zonas promisorias del espacio de búsqueda. Emplea, además, un proceso de contracción que se encarga se seleccionar los mejores representantes de cada zona explorada. Nuestra propuesta es aprovechar las fortalezas del algoritmo y adaptarlo al plegamiento de proteínas. Después de los experimentos realizados se concluye que sí es posible aplicar la meta-heurística VMO para predecir dichas estructura. La no existencia de una función objetivo capaz de cumplir la hipótesis termodinámica del plegamiento, imposibilitó el logro de mejores resultados.
Finalmente, se comentan algunas ideas para mejorar el método propuesto, también se realizan varias propuestas de posibles líneas de investigación a acometer en el futuro.
This work describes the protein folding which constitutes one of the most important challenges in molecular biology. Due to the fact that this problem is extremely difficult, there is not method known yet capable of resolving it efficiently. Therefore this report proposes an optimization algorithm based on Variable Grids (VMO) to address this problem. The algorithm arises as a population model conceived by the Artificial Intelligence group at our university. The VMO model uses an expansion process consisting of generation functions for new solutions that explore promising areas of searching space. It also uses a contraction process responsible for selecting the best representatives of each area explored. Our proposal is to utilize the strengths of the algorithm and adapt it to protein folding. After the performed experiments it is concluded that it is possible to apply the heuristic goal VMO to predict these structures. The lack of an objective function capable of fulfilling the thermodynamic hypothesis of folding rendered it impossible to achieve better results. Finally, we discuss some ideas for improving the proposed method; also several suggestions are made for possible research lines to undertake in the future.
This work describes the protein folding which constitutes one of the most important challenges in molecular biology. Due to the fact that this problem is extremely difficult, there is not method known yet capable of resolving it efficiently. Therefore this report proposes an optimization algorithm based on Variable Grids (VMO) to address this problem. The algorithm arises as a population model conceived by the Artificial Intelligence group at our university. The VMO model uses an expansion process consisting of generation functions for new solutions that explore promising areas of searching space. It also uses a contraction process responsible for selecting the best representatives of each area explored. Our proposal is to utilize the strengths of the algorithm and adapt it to protein folding. After the performed experiments it is concluded that it is possible to apply the heuristic goal VMO to predict these structures. The lack of an objective function capable of fulfilling the thermodynamic hypothesis of folding rendered it impossible to achieve better results. Finally, we discuss some ideas for improving the proposed method; also several suggestions are made for possible research lines to undertake in the future.
Descripción
Palabras clave
Optimización, Mallas Variables, Predicción, Plegamiento de Proteínas, Bioinformática, Inteligencia Artificial