Neural network-based method for peptide identification in proteomics

Abstract

Protein identification in biological samples is one of the main objectives of proteomics. In proteomic experiments proteins are first digested into short peptides, which are next analyzed using tandem mass spectrometry and identified by database search algorithms. In this study a novel neural network-based method for peptide identification is proposed. The presented method improves the identification efficiency by the incorporation of additoinal peptide-specific features and scores from multiple database search algorithms. Moreover, the method for filtering out low quality mass spectra prior to database search in order to reduce the overall computational time of the identification process is presented. © 2012 Springer-Verlag.