Automatic determination of O-glycan structure from fragmentation spectra.

Glycosylation is one of the most important classes of post-translational protein modifications, but the identification of glycans is difficult because of their branched structures and numerous isomers. We describe an algorithm called CartoonistTwo that proposes structures for O-linked glycans by automatically analyzing fragmentation mass spectra. CartoonistTwo improves upon previous glycan identification software primarily in its scoring function, which can more successfully distinguish among a number of similar structures. CartoonistTwo was designed and tested with FTICR mass spectra, and includes automatic recalibration and peak selection especially tuned for such data, yet it can be easily adapted to fragmentation spectra (MS2 or MSn) from other instrument types. On a validated test set of 34 SORI-CID MSn FTICR spectra from Xenopus egg jelly, CartoonistTwo gave the manually determined structural assignment either the first or second highest score over 90% of the time. And for over 50% of these spectra, CartoonistTwo selected a unique highest scoring structure that agreed with the manually determined one.