Multiattribute Glycan Identification and FDR Control for Glycoproteomics

Abstract

Rapidly improving methods for glycoproteomics have enabled increasingly large-scale analyses of complex glycopeptide samples, but annotating the resulting mass spectrometry data with high confidence remains a major bottleneck. We recently introduced a fast and sensitive glycoproteomics search method in our MSFragger search engine, which reports glycopeptides as a combination of a peptide sequence and the mass of the attached glycan. In samples with complex glycosylation patterns, converting this mass to a specific glycan composition is not straightforward; however, as many glycans have similar or identical masses. Here, we have developed a new method for determining the glycan composition of N-linked glycopeptides fragmented by collisional or hybrid activation that uses multiple sources of information from the spectrum, including observed glycan B-type (oxonium) and Y-type ions and mass and precursor monoisotopic selection errors to discriminate between possible glycan candidates. Combined with false discovery rate estimation for the glycan assignment, we show that this method is capable of specifically and sensitively identifying glycans in complex glycopeptide analyses and effectively controls the rate of false glycan assignments. The new method has been incorporated into the PTM-Shepherd modification analysis tool to work directly with the MSFragger glyco search in the FragPipe graphical user interface, providing a complete computational pipeline for annotation of N-glycopeptide spectra with false discovery rate control of both peptide and glycan components that is both sensitive and robust against false identifications.

Keywords: false discovery rate; glycoproteomics; glycosylation; mass spectrometry; software.

Graphical abstract

Fig. 1

Glycan assignment workflow in PTM-Shepherd.A, glyco search in MSFragger and peptide FDR filtering in Philosopher annotates a spectrum with a peptide and a delta mass. B, possible glycan candidates with masses similar to the provided delta mass are gathered from the internal glycan list in PTM-Shepherd. C, pairwise comparison of all candidates determines the best match to the spectrum based on unique fragment ions for each candidate and mass and isotope errors. Note that mass error is calculated after first correcting any isotope error. Green check marks in a candidate's column indicate positive evidence for that candidate, red and orange x's indicate negative or slightly negative evidence, respectively. D, the best candidate is rescored to generate the absolute score, and FDR is computed using the score distribution of target and decoy glycans. FDR, false discovery rate.

Fig. 2

Results of entrapment searches of yeast dataset.A, Venn diagram of the two glycan lists used in MSFragger searches. B, GlycoPSMs annotated at 1% peptide FDR (blue) and combined 1% peptide and glycan FDR (green) for MSFragger searches of the glycan lists with PTM-Shepherd glycan FDR filtering. ∗Indicates ammonium adduction was allowed. Comparison with pGlyco 3 reported results for the same peptide database, glycan list, and FDR levels is at right either with or without allowing one ammonium adduct per glycan (encoded by pGlyco3 as “aH” residues). C, unique glycoproteins, glycopeptide sequences, and glycan–peptide combinations (each glycan composition on each unique peptide sequence counts as a separate entry) detected in MSFragger + PTM-Shepherd (green) and pGlyco3 (orange) results at 1% peptide and glycan FDR. FDR, false discovery rate; PSM, peptide-spectrum match.

Fig. 3

Impact of individual score components on glycan assignment performance.A, glycoPSMs passing 1% glycan (and peptide) FDR with full score (all components) or one component removed. B, table of total glycoPSMs and entrapment glycoPSMs of various types for each analysis presented in A. FDR, false discovery rate; PSM, peptide-spectrum match.

Fig. 4

Comparison of glycoPSMs annotated from mouse brain tissue from Riley et al.A, glycoPSMs annotated by the original analysis in Byonic and our reanalysis with MSFragger and PTM-Shepherd. The improvement in annotated spectra from MSFragger search is maintained after applying 1% glycan FDR filtering. B, unique glycoproteins, glycopeptide sequences, and glycan–peptide sequence combinations reported by MSFragger/PTM-Shepherd or Riley et al. C, glycan composition categories observed sorted by residue type(s) contained, showing few unexpected (NeuGc or sulfate-containing) glycans. D, analysis of paired HCD and AI-ETD scans of the same precursor from MSFragger/PTM-Shepherd search, assessed for whether the same peptide and glycan were matched in both of the paired scans. AI–ETD, activated ion–electron transfer dissociation; FDR, false discovery rate; HCD, higher energy collisional dissociation; PSM, peptide-spectrum match.

Fig. 5

Comparison of N-glycan search in mouse tissue dataset between MSFragger + PTM-Shepherd and pGlyco3.A, glycoPSMs passing 1% peptide and glycan FDR from MSFragger/PTM-Shepherd, pGlyco3, or pGlyco2 searches in each of five tissue types. B, comparison of the number of unique glycoproteins and glycopeptide sequences (1% peptide and glycan FDR) from searches presented in A. C, PTM-Shepherd assigned compositions by residues included in the glycan. Note that glycoPSMs containing multiple residue types (e.g., fucose and NeuAc) will be counted in multiple categories. D, comparison of PTM-Shepherd assigned glycan compositions with pGlyco3 compositions. Positive values indicate more of a residue type assigned by PTM-Shepherd, and negative values indicate more of a residue type assigned by pGlyco3. In all tissues, PTM-Shepherd assigns more sialic acid compositions, whereas pGlyco3 assigned more fucose. FDR, false discovery rate; PSM, peptide-spectrum match.