Fig. 1: Workflow of GlyPep-Quant.

The quantitative analysis in GlyPep-Quant includes two modules: label-free quantification with match-between-runs (LFQ-MBR) (left panel, red) and library-based identification and quantification (library-based virtual MBR, right panel, blue). In the LFQ-MBR process, MS files I and II in dataset A are utilized. Elution profiles are firstly extracted based on GPSM identification in MS file I, dataset A, and then the matched elution profiles are extracted from the matched data (MS file II, dataset A) using a two-step elution extraction method. The matched elution profiles include target (same precursor m/z value) and decoy elution profiles, and their confidence is evaluated by a well-trained machine learning model. For the generation of decoy elution profile, a random mass shift of 3–20 Th is applied to the original precursor ion m/z value, and the decoy elution profile is extracted within the same retention time range as the original. In library-based identification and quantification, an MS1 feature library is firstly constructed by summarizing MS1 features in a large dataset A acquired under the same conditions for a large cohort of samples, it is then applied to identify and quantify glycopeptides for a newly acquired MS file (or files) (from new dataset B) via a dedicated library-based virtual match-between-runs algorithm. LC-MS/MS runs in dataset B are newly acquired using the same sample types and experimental conditions as dataset A. Target and decoy elution profiles in the MS file(s) are extracted and matched based on the MS1 feature library. The decoy profile generation method is the same as MBR. The confidence of the matched result is evaluated by a separate machine learning model. Based on the quantitative results, a new biomarker discovery strategy is developed, which enables the screening of the site-specific glycan abundance ratios and allows diagnosis of individuals with newly acquired MS data.