Herramienta computacional DASDE® para aplicaciones quimio-bioinformáticas usando algoritmos genéticos en la construcción de multiclasificadores
Fecha
2016-06-20
Autores
Peña Díaz, Claudia
Título de la revista
ISSN de la revista
Título del volumen
Editor
Universidad Central “Marta Abreu” de Las Villas
Resumen
El descubrimiento de un nuevo medicamento y desarrollo posterior del mismo, son dos fases que comprenden alrededor de 12 años de investigación en países desarrollados, con un costo promedio de 231 millones de dólares por cada nuevo medicamento que sale al mercado. El desarrollo de la ciencia computacional, ha permitido la búsqueda racional de nuevas alternativas terapéuticas, siendo los métodos in silico una de las técnicas que mejoran potencialmente el descubrimiento y posterior desarrollo de fármacos. La utilización de multiclasificadores junto con algoritmos de búsqueda heurística son estrategias, que ayudan en el pronóstico del análisis y a disminuir su tiempo para obtener buenos resultados en la búsqueda y posterior desarrollo de compuestos activos frente a enfermedades prioritarias a nivel mundial. El software DASDE v1.0 desarrollado en el año 2013 para investigadores de perfil químico de nuestra universidad tiene varias funcionalidades, dentro de ellas permite la selección de modelos diversos para obtener posteriormente, combinaciones de los mismos donde se mejoren los resultados alcanzados individualmente, estas funcionalidades se realizan de manera exhaustiva lo que conlleva a grandes demoras en las ejecuciones puesto que las aplicaciones en el campo de la quimio-bioinformática siempre cuentan con un gran número de clasificadores y por lo tanto la cantidad de combinaciones que se pueden generar es considerablemente grande. En este trabajo se propone la incorporación al software de la meta heurística Algoritmos Genéticos para resolver el problema combinatorio que se presenta y minimizar el tiempo de búsqueda de las soluciones, garantizándole al usuario obtener combinaciones que sean igualmente buenas en calidad. Finalmente se presenta el análisis con dos bases de datos reales en el campo de la quimio-bioinformática.
The discovery of a new drug and subsequent development are two phases comprising about 12 years of research in developed countries, with an average cost of $ 231 million per new drug hitting the market. The development of computational science, has allowed the rational search for new therapeutic alternatives, where in silico methods are one of the techniques that potentially improve the discovery and development of drugs. The use of multi classifiers with heuristic search algorithms are strategies that help in forecasting the analysis and decrease time to get good results in the search and subsequent development of active compounds against priority diseases worldwide. DASDE v1.0 software, developed in 2013 for researchers with a chemical profile of our university, has several features, among them it´s allowed the selection of diverse models for later combinations where the results achieved improve individual ones. These features are performed exhaustively which leads to long delays in executions since the applications in the field of chemo-bio-informatics always have a large number of classifiers and therefore the number of combinations that can be generated is considerably large. In this paper the incorporation of meta heuristic genetic algorithms to the software is proposed to solve the combinatorial problem that occurs and to minimize search time of solutions, guaranteeing the user to obtain combinations that are equally good in quality. Finally, the analysis with two real databases in the field of chemo-bio-informatics is presented.
The discovery of a new drug and subsequent development are two phases comprising about 12 years of research in developed countries, with an average cost of $ 231 million per new drug hitting the market. The development of computational science, has allowed the rational search for new therapeutic alternatives, where in silico methods are one of the techniques that potentially improve the discovery and development of drugs. The use of multi classifiers with heuristic search algorithms are strategies that help in forecasting the analysis and decrease time to get good results in the search and subsequent development of active compounds against priority diseases worldwide. DASDE v1.0 software, developed in 2013 for researchers with a chemical profile of our university, has several features, among them it´s allowed the selection of diverse models for later combinations where the results achieved improve individual ones. These features are performed exhaustively which leads to long delays in executions since the applications in the field of chemo-bio-informatics always have a large number of classifiers and therefore the number of combinations that can be generated is considerably large. In this paper the incorporation of meta heuristic genetic algorithms to the software is proposed to solve the combinatorial problem that occurs and to minimize search time of solutions, guaranteeing the user to obtain combinations that are equally good in quality. Finally, the analysis with two real databases in the field of chemo-bio-informatics is presented.
Descripción
Palabras clave
Herramientas Computacionales, Quimiobioinformática, Aplicaciones, Algoritmos Genéticos, Construcción de Multiclasificadores