Site-specific identification of phosphorylation is one of the main computational challenges in phosphoproteomics. Earlier we proposed a method based on simulated phosphopeptide spectral libraries  that together with our software, SimPhospho , facilitates highly sensitive and accurate phosphosite assignments. Here, we will investigate how well multiply phosphorylated peptides can be simulated.
The principle of the simulation algorithm is as follows. Based on the spectra of dephosphorylated or nonmodified peptides, we predict spectra of phosphorylated peptides using the information about present ion peaks and their intensities as reference. Different types of ions, as well as their 2+, 3+ charged ions, and neutral losses are taken into account. Simulated spectra are collected in spectral libraries and used for identification of MS/MS spectra of phosphopeptide samples using SpectraST. For method development, hundred phosphopeptides (singly-, doubly- and triply-phosphorylated) were selected to be synthesized. These peptides were divided into three pools making sure that the phosphopeptide isoforms were separated. Analysis was done by Orbitrap HCD in Thermo Scientific Q Exactive mass spectrometer.
Results and discussion
The earlier version of SimPhospho was biased towards detection of singly phosphorylated peptides. This limitation is addressed by implementing simulation of doubly- and triply-phosphorylated peptides in SimPhospho 2. We will compare various combinations of simulation parameters in terms of False Localization Rate of phosphorylation sites we are able to obtain, when performing spectral library search of samples with known phosphorylation sites.