Generalized Approach for Determination of False Localization Rates in Post-Translational Modification Site Assignments

摘要

Our method for PTM site assignment provides two major contributions. The first is a robust way to generate decoy hits for calculation of false localization rates for a set of modified data. This method can be applied with any scoring scheme which independently assigns scores to every possible residue on a peptide. With our method for calculating FLR, we now can adjust and compare scoring methods to determine directly how those choices affect the overall false discovery rate for site localization. The second contribution is a scoring method which can handle site localization ambiguity as well as mixture spectra. In our phosphorylation data, for example, there is a large amount of ambiguity in site assignment, some of it stemming from water losses not related to the loss of the phosphate. Also, in the case of histones, the ability to handle ambiguity and multiple possible modification variants contributing to a spectrum will facilitate the deconvolution of the modified peptide spectra. While site localization remains an open problem, we are confident that experiment-wide methods for calculation of False Discovery Rates for PTM site assignment, such as the False Localization Rate approach proposed here, will be essential to enable robust and comparable reporting of PTM site assignments.