D-score: a search engine independent MD-score.

While peptides carrying PTMs are routinely identified in gel-free MS, the localization of the PTMs onto the peptide sequences remains challenging. Search engine scores of secondary peptide matches have been used in different approaches in order to infer the quality of site inference, by penalizing the localization whenever the search engine similarly scored two candidate peptides with different site assignments. In the present work, we show how the estimation of posterior error probabilities for peptide candidates allows the estimation of a PTM score called the D-score, for multiple search engine studies. We demonstrate the applicability of this score to three popular search engines: Mascot, OMSSA, and X!Tandem, and evaluate its performance using an already published high resolution data set of synthetic phosphopeptides. For those peptides with phosphorylation site inference uncertainty, the number of spectrum matches with correctly localized phosphorylation increased by up to 25.7% when compared to using Mascot alone, although the actual increase depended on the fragmentation method used. Since this method relies only on search engine scores, it can be readily applied to the scoring of the localization of virtually any modification at no additional experimental or in silico cost.

A complex standard for protein identification, designed by evolution.

Shotgun proteomic investigations rely on the algorithmic assignment of mass spectra to peptides. The quality of these matches is therefore a cornerstone in the analysis and has been the subject of numerous recent developments. In order to establish the benefits of novel algorithms, they are applied to reference samples of known content. However, these were recently shown to be either too simple to resemble typical real-life samples or as leading to results of lower accuracy as the method itself. Here, we describe how to use the proteome of Pyrococcus furiosus , a hyperthermophile, as a standard to evaluate proteomics identification workflows. Indeed, we prove that the Pyrococcus furiosus proteome provides a valid method for detecting random hits, comparable to the decoy databases currently in popular use, but we also prove that the Pyrococcus furiosus proteome goes squarely beyond the decoy approach by also providing many hundreds of highly reliable true positive hits. Searching the Pyrococcus furiosus proteome can thus be used as a unique test that provides the ability to reliably detect both false positives as well as proteome-scale true positives, allowing the rigorous testing of identification algorithms at the peptide and protein level.