Diseño y creación de un clasificador de péptidos antibacterianos utilizando técnicas de procesamiento digital de señales y algoritmos de aprendizaje
Este informe presenta la descripción del diseño y la obtención de un clasificador de péptidos antimicrobianos que permite realizar una preselección de cadenas peptídicas, de manera que se reduzca el número de experimentos necesarios para encontrar alguna con actividad antibacteriana. Se hace uso de...
- Autores:
-
Coba Cruz, Diego Fernando
- Tipo de recurso:
- http://purl.org/coar/version/c_b1a7d7d4d402bcce
- Fecha de publicación:
- 2016
- Institución:
- Universidad Industrial de Santander
- Repositorio:
- Repositorio UIS
- Idioma:
- spa
- OAI Identifier:
- oai:noesis.uis.edu.co:20.500.14071/35074
- Palabra clave:
- Clasificador
Péptido
Máquinas De Soporte Vectorial
Validación Cruzada
Knn
Potencial De Interacción Ion Electrón
Señal Discreta
Representación En Frecuencia.
This paper presents the design and implementation of an antibacterial peptides classifier that allow the pre-selection of those sequences that possess antibacterial activity in order to reduce the quantity of experiments that should be performed to find a successful antibacterial peptide. Supervised learning models were used in order to teach the system how to discriminate a peptide with Support vector machines with linear and radial basis functions
and k-nearest neighbors with uniform and distance dependent weights were implemented. Vector and frequency spectrum mathematical representations where used both normalized and no normalized forms. For each combination of learning model
learning model parameters and mathematical representation
the estimated assessment was determined through Nested K Fold Cross Validation process to finally take the one with the best performance which was obtained using K Fold Cross Validation process. Finally
a comparison between the different learning models and representations was made as conclusions of this work. The software created for this purpose is made by modules that allow the development of each one of the process stages. The final result is composed of the obtained classifiers
their estimated assessment metrics file and the system through which were gotten. 3
- Rights
- License
- Attribution-NonCommercial 4.0 International (CC BY-NC 4.0)