Protein glycosylation plays a significant role in numerous physiological and pathological cellular functions. Glycoproteomics based on liquid chromatography-tandem mass spectrometry (LC-MS/MS) studies the protein glycosylation on a proteome-wide scale to get combinational information on glycosylation site, glycosylation level and glycan structure. However, the current database searching-based methods for glycoproteomics often fall short in glycan structure determination due to the limited occurrence of structure-determining ions. While spectral searching methods can utilize fragment intensity information to facilitate the identification of glycopeptides, its application is hindered by the difficulties in spectral library construction. In this work, we present DeepGP, a hybrid deep learning framework based on Transformer and graph neural network (GNN), for the prediction of MS/MS spectra and retention time of glycopeptides. Two GNN modules are utilized to capture the branched glycan structure and predict glycan ions intensity respectively. Additionally, a pre-training strategy is implemented to alleviate the insufficiency of glycoproteomics data. Testing on multiple biological datasets, we demonstrate that DeepGP can predict MS/MS spectra and retention time of glycopeptides closely aligning with the experimental results. Comprehensive benchmarking of DeepGP on synthetic and biological datasets validates its effectiveness in distinguishing similar glycoforms . Remarkably, DeepGP can differentiate isomeric glycopeptides using MS/MS spectra without diagnostic ions. Based on various decoy methods, we demonstrated that DeepGP in combination with database searching can significantly increase the detection sensitivity of glycopeptides. We outlook that DeepGP can inspire extensive future work in glycoproteomics.