fmpRPMF: A Web Implementation for Protein Identification by Robust Peptide Mass Fingerprinting

Peptide mass fingerprinting continues to play an important role in current proteomics studies based on its good performance in sample throughput, specificity for single peptides, and insensitivity to unexpected post-translational modifications as compared with MS n . We previously proposed and evaluated the use of feature-matching pattern-based support vector machines (SVMs) for robust protein identification. This approach is now facilitated with an updated web server (fmpRPMF) incorporated with several newly developed or improved modules and workflows allowing identification of proteins from MS 1 data. Development of the latest fmpRPMF web tool successfully provides a rapid and effective strategy for narrowing the range of candidate proteins. First, a mass-scanning procedure screens all candidate proteins matching the theoretical peptide mass at least three times, thereby reducing the number of candidate proteins from tens of thousands to thousands. Second, a crude ranking procedure screens true-positive proteins among the top six ranked times of candidates based on 17 selected features to reduce the number used for SVM prediction from thousands to tens. The improvement of forecasting efficiency met the requirements of multi-user and multi-task identification for web services. The updated fmpRPMF server is freely available at http://bioinformatics.datawisdom.net/fmp .