SpecGlob: rapid and accurate alignment of mass spectra differing from their peptide models by several unknown modifications

BackgroundIn proteomics, mass spectra representing peptides carrying multiple unknown modifications are particularly difficult to interpret, which results in a large number of unidentified spectra. MethodsWe developed SpecGlob, a dynamic programming algorithm that aligns pairs of spectra, each such pair being a Peptide-Spectrum Match (PSM) provided by any Open Modification Search (OMS) method. For each PSM, SpecGlob computes the best alignment according to a given score system, while interpreting the mass delta within the PSM as one or several unspecified modification(s). All alignments are provided in a file, written in a specific syntax. ResultsUsing several sets of simulated spectra generated from the human proteome, we demonstrate that running SpecGlob as a post-analysis of an OMS method can significantly increase the number of correctly interpreted spectra, as SpecGlob is able to correctly and rapidly align spectra that differ by one or more modification(s) without any a priori. ConclusionSince SpecGlob explores all possible alignments that may explain the mass delta within a PSM, it reduces interpretation errors generated by incorrect assumptions about the modifications present in the sample or the number and the specificities of modifications carried by peptides. Our results demonstrate that SpecGlob should be relevant to align experimental spectra, although this consists in a more challenging task.