Unrestricted identification of post translational modifications from tandem mass spectra datasets

Identification of post-translational modifications (PTMs) is an important task for understanding biological functions in proteomics. From tandem mass spectra, the identification algorithms attempt to discover accurate PTMs with short time. However, spectral imperfection, such as noise peaks and missing peaks, in tandem mass spectra provokes computational artifacts and often hampers the identification of PTM. In order to address this problem, we propose an unrestricted PTM identification algorithm that processes stepwise complete mass shift search which decreases the computational complexity and increases the performance by filtering computational artifacts. The proposed algorithm detects, clusters, and evaluates all mass changes on multiple sites in candidate peptides. The optimal combinations of top-scoring mass shifts are scored with matched peaks and assigned to PTMs in Unimod. In a test with simulated spectra using different missing peak ratios, our algorithm showed reasonable accuracy with missing peak ratio up to 70%. It is robust against noise and missing of precursor ion mass. In a test with HUPO Brain Proteome Project (BPP) datasets, the total coverage of the search results against BPP annotation was 95.38%. The performance of identifying multiple PTMs in various spectral conditions is substantially higher than in previous methods.