Improving Confidence for the Identification of N-Linked Intact Glycopeptides Based on the Precursor Feature

As a common and heterogeneous post-translational modification (PTM), glycosylation takes part in a wide variety of biological processes. Precision analysis of glycosylation is of great value for the determination of their functional roles and the discovery of novel disease biomarkers. Mass spectrometry-based identification of intact glycopeptide has become increasingly popular in glycosylation studies due to its ability for high-throughput analysis of glyco-sites and glycan modifications simultaneously. Variable modification searching strategy by setting database glycans as variable modification on glycosites is commonly used in the identification of intact glycopeptides. Due to the issues of search space and random match, the performance of variable modification strategy is affected by the size of glycan database adopted in analysis. Inappropriate using of glycan databases in intact glycopeptide identification often results in poor analysis coverage or high false positive rate. In addition, the misassignment of precursor, which causes incorrect derivation of mass value of glycopeptides, is also frequently observed in the identification process and leads to incorrect glycopeptide spectrum interpretation. As only peptide fragment ions in glycopeptide spectra are matched by setting the glycan part as variable modifications, the confidence of precursor assignment and glycopeptide identification can not be fully assessed by spectrum matching scores, which are often solely based on peptide fragment ions. The spectrum features of glycopeptide precursor, including elution profiles and intensity distributions of isotopic peak list, can be utilized to screen the incorrect glycopeptide results caused by isotopic shift, which are great helpful to elevate the reliability of identification results. In this work, the effect of glycan databases on the performance of variable modification searching was firstly investigated by using the dataset of yeast as a benchmark, since only oligo-mannose N-glycans were synthesized in yeast. Applying the method of precursor feature screening, the false discovery rate of glycopeptide identification was reduced from 11.31% to 4.15% on the dataset of yeast. Further analysis for the glycan part of glycopeptide results proved that precursor feature screening was able to remove false positive identifications at high specificity. In addition, investigation on the dataset of mouse tissues also indicated that the precursor feature screening method could be utilized to improve the identification confidence for the analysis of complex samples. Being different from fragment ion matching, precursor feature screening method provides an alternative approach to assess the confidence of glycopeptide identification by utilizing the information of precursor isotopic peaks, which is of potential helpful for glycosylation studies.