Predicting peroxidase subcellular location by hybridizing different descriptors of Chou' pseudo amino acid patterns.
Anal Biochem
; 458: 14-9, 2014 Aug 01.
Article
em En
| MEDLINE
| ID: mdl-24802134
ABSTRACT
Peroxidases as universal enzymes are essential for the regulation of reactive oxygen species levels and play major roles in both disease prevention and human pathologies. Automated prediction of functional protein localization is rarely reported and also is important for designing new drugs and drug targets. In this study, we first propose a support vector machine (SVM)-based method to predict peroxidase subcellular localization. Various Chou' pseudo amino acid descriptors and gene ontology (GO)-homology patterns were selected as input features to multiclass SVM. Prediction results showed that the smoothed PSSM encoding pattern performed better than the other approaches. The best overall prediction accuracy was 87.0% in a jackknife test using a PSSM profile of pattern with width=5. We also demonstrate that the present GO annotation is far from complete or deep enough for annotating proteins with a specific function.
Palavras-chave
Texto completo:
1
Coleções:
01-internacional
Base de dados:
MEDLINE
Assunto principal:
Peroxidase
/
Máquina de Vetores de Suporte
Tipo de estudo:
Prognostic_studies
/
Risk_factors_studies
Limite:
Humans
Idioma:
En
Revista:
Anal Biochem
Ano de publicação:
2014
Tipo de documento:
Article