The knowledge of the cellular location of a protein is a key step towards understanding its function. The availability of systems that can predict location from a protein sequence will be essential to the full characterization of expressed proteins. In this paper, we show that the performances of Subspace classifier are slightly superior to the performance of Support vector machine. Morover, we show that the performances of an ensemble of subspace classifiers are superior to the performance of a “stand-alone” subspace classifier. The Reinhardt and Hubbard database has been used to examine the performance of the multiclassifier method. The rates of correct prediction for the three possible subcellular location of prokaryotic proteins are 91.2%. The rates of correct prediction for the four possible subcellular location of eukaryotic proteins are 81.2%. Predictions by our approach are robust to errors in the protein N-terminal sequences.
A. Lumini, L. Nanni (2005). Subspace classifier for protein subcellular localization prediction. INNSBRUCK : ACTA press.
Subspace classifier for protein subcellular localization prediction
LUMINI, ALESSANDRA;NANNI, LORIS
2005
Abstract
The knowledge of the cellular location of a protein is a key step towards understanding its function. The availability of systems that can predict location from a protein sequence will be essential to the full characterization of expressed proteins. In this paper, we show that the performances of Subspace classifier are slightly superior to the performance of Support vector machine. Morover, we show that the performances of an ensemble of subspace classifiers are superior to the performance of a “stand-alone” subspace classifier. The Reinhardt and Hubbard database has been used to examine the performance of the multiclassifier method. The rates of correct prediction for the three possible subcellular location of prokaryotic proteins are 91.2%. The rates of correct prediction for the four possible subcellular location of eukaryotic proteins are 81.2%. Predictions by our approach are robust to errors in the protein N-terminal sequences.I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.