Estudios QSAR de la actividad antibacterial usando el método TOMOCOMD-CARDD
Archivos
Fecha
2006-06-25
Autores
Kaloko, Amirou
Título de la revista
ISSN de la revista
Título del volumen
Editor
Universidad Central "Marta Abreu" de las Villas
Resumen
El objetivo fundamental de este estudio fue desarrollar simples relaciones cuantitativas
estructura-actividad (QSAR) para la clasificación y la predicción de la actividad antibacteriana, de manera que permita el desarrollo de procesos de “screening” in silico. Con este fin, una data de 2030 compuestos, clasificados acorde a su actividad antibacteriana, y un total de cuatro familias de descriptores TOMOCOMD-CARDD, fueron calculados y analizados. Para identificar los descriptores que permitan la separación en dos clases (ej: compuestos con y sin actividad antibacteriana), el análisis de “pasos-hacia-delante” fue empleado como métodos de selección de
variables; y entonces los modelos fueron desarrollados usando el análisis discriminante lineal.
Para acceder al poder predictivo del modelo, se realizó una validación con una serie de predicción externa. Los resultados de los análisis indican que los descriptores TOMOCOMD-CARDD totales y locales (heteroátomos e H-unidos a heteroátomos), proporcionan una excelente separación de la data (> 90 y >89%) en la serie de entrenamiento y en la serie de predicción, respectivamente. Los modelos desarrollados fueron usados en una búsqueda virtual de compuestos antibacterianos tipo-fármacos; >87% de los compuestos empleados en esta simulación fueron correctamente clasificados, los cual es indicativo de la capacidad de los modelos TOMOCOMD-CARDD en el descubrimiento de nuevos compuestos líderes, desde el
punto de vista estructural y de modo de acción. Finalmente, los modelos QSAR obtenidos pueden ser aplicados a grandes bibliotecas virtuales con el objetivo de descubrir/seleccionar compuestos candidatos como antibacterianos.
The aim of this study was to develop a simple quantitative structure-activity relationship (QSAR) for the classification and prediction of antibacterial activity, so as to enable in silico screening. To this end a database of 2030 compounds, classified according to whether they had antibacterial activity, and for which a total of four TOMOCOMD-CARDD descriptors´s families were calculated, was analyzed. To identify descriptors that allowed separation of the two classes (i.e. those compounds with and without antibacterial activity), analysis of forward stepwise was utilized like variable selection´s method, and models were developed using linear discriminant analysis. Model predictivity was assessed and validated by the used of an external test set, for which predictions were made from the model. The results of the analyses indicated that total and local (heteroatoms and H-bonding heteroatoms) TOMOCOMD-CARDD descriptors, provide excellent separation of the data (>90% y >89% in training and test set, respectively). The models developed are then used in a simulation of virtual search of antibacterial drug-like compounds; >87% of the chemicals used in this simulated search were correctly classified, thus indicating the ability of the TOMOCOMD-CARDD models of finding lead compounds with novel structures and action mode. So, the obtained QSAR model can be applied to a large set of compounds searching for new candidates as antibacterials.
The aim of this study was to develop a simple quantitative structure-activity relationship (QSAR) for the classification and prediction of antibacterial activity, so as to enable in silico screening. To this end a database of 2030 compounds, classified according to whether they had antibacterial activity, and for which a total of four TOMOCOMD-CARDD descriptors´s families were calculated, was analyzed. To identify descriptors that allowed separation of the two classes (i.e. those compounds with and without antibacterial activity), analysis of forward stepwise was utilized like variable selection´s method, and models were developed using linear discriminant analysis. Model predictivity was assessed and validated by the used of an external test set, for which predictions were made from the model. The results of the analyses indicated that total and local (heteroatoms and H-bonding heteroatoms) TOMOCOMD-CARDD descriptors, provide excellent separation of the data (>90% y >89% in training and test set, respectively). The models developed are then used in a simulation of virtual search of antibacterial drug-like compounds; >87% of the chemicals used in this simulated search were correctly classified, thus indicating the ability of the TOMOCOMD-CARDD models of finding lead compounds with novel structures and action mode. So, the obtained QSAR model can be applied to a large set of compounds searching for new candidates as antibacterials.
Descripción
Palabras clave
Actividad Antibacteriana, Antibacterial Activity, TOMOCOMD-CARDD, Bibliotecas Virtuales, Diseño de Fármacos