PhosphoHunter: An Efficient Software Tool for Phosphopeptide Identification

Abstract

Phosphorylation is a protein posttranslational modification. It is responsible of the activation/inactivation of disease-related pathways, thanks to its role of "molecular switch." The study of phosphorylated proteins becomes a key point for the proteomic analyses focused on the identification of diagnostic/therapeutic targets. Liquid chromatography coupled to tandem mass spectrometry (LC-MS/MS) is the most widely used analytical approach. Although unmodified peptides are automatically identified by consolidated algorithms, phosphopeptides still require automated tools to avoid time-consuming manual interpretation. To improve phosphopeptide identification efficiency, a novel procedure was developed and implemented in a Perl/C tool called PhosphoHunter, here proposed and evaluated. It includes a preliminary heuristic step for filtering out the MS/MS spectra produced by nonphosphorylated peptides before sequence identification. A method to assess the statistical significance of identified phosphopeptides was also formulated. PhosphoHunter performance was tested on a dataset of 1500 MS/MS spectra and it was compared with two other tools: Mascot and Inspect. Comparisons demonstrated that a strong point of PhosphoHunter is sensitivity, suggesting that it is able to identify real phosphopeptides with superior performance. Performance indexes depend on a single parameter (intensity threshold) that users can tune according to the study aim. All the three tools localized >90% of phosphosites.

Figure 1

Summary of PhosphoHunter procedure. Block A (implemented via the create_database.pl script): a database in FASTA format is used to create a target database according to appropriate digestion rules and other parameters provided in an input file, such as the number of allowed consecutive missing cleavages. The database is then used to obtain a decoy database and a single composite database is obtained from target and decoy databases. Block B (implemented via the merge.pl script): individual dta files, corresponding to experimental spectra, are merged into a single dta file. Block C: experimental spectra are normalized and processed by discarding charges higher than 4, low-intensity peaks, and peptides not showing neutral loss. The intensity threshold of neutral loss is specified in the input file. Block D: theoretical and processed spectra are compared according to a scoring function and a list of phosphopeptides with scores is associated with each spectrum. Block E: for each spectrum, a p-value is computed for each element of the list and only the peptides with a p-value below a specific threshold, defined in the input file (relation not shown in the figure), are kept in the final list. Blocks C, D, and E are all implemented via the phosphopeptide_ID.pl script.