Relative Retention Time Estimation Improves N‑Glycopeptide Identifications by LC–MS/MS

Glycopeptides identified
by tandem mass spectrometry rely on the
identification of the peptide backbone sequence and the attached glycan(s)
by the incomplete fragmentation of both moieties. This may lead to
ambiguous identifications where multiple structures could explain
the same spectrum equally well due to missing information in the mass
spectrum or incorrect precursor mass determination. To date, approaches
to solving these problems have been limited, and few inroads have
been made to address these issues. We present a technique to address
some of these challenges and demonstrate it on previously published
data sets. We use a linear modeling approach to learn the influence
of the glycan composition on the retention time of a glycopeptide
and use these models to validate glycopeptides within the same experiment,
detecting over 400 incorrect cases during the MS/MS search and correcting
75 cases that could not be identified based on mass alone. We make
this technique available as a command line executable program, written
in Python and C, freely available at https://github.com/mobiusklein/glycresoft in source form, with precompiled binaries for Windows.