Sweet artificial intelligence as digital catalysts in infectious diseases: Glycomics and glycoanalysis

Glycans mediate microbial adhesion, host-cell entry, immune recognition and immune evasion, but their branching topology, linkage isomerism and context-dependent functions make glycan-centered infectious-disease research difficult to standardize. This review summarizes how artificial intelligence (AI), machine learning (ML) and deep learning (DL) are being applied to glycomics and glycoanalysis in infection biology. We first outline major sample sources, analytical platforms and curated databases used to generate glycan data, including mass spectrometry (MS), nuclear magnetic resonance (NMR), chromatography and glycan/lectin microarrays. We then compare supervised learning, unsupervised learning, graph neural networks(GNNs), sequence models and transformer-informed frameworks for glycan structure analysis, lectin-glycan binding prediction, pathogenicity assessment and biomarker discovery. Representative validated models, including SweetNet and LectinOracle, demonstrate that glycan-aware AI can achieve reproducible predictive performance across independent datasets. Finally, we discuss diagnostic applications, virulence forecasting, vaccine and drug development, and pathogen-related autoimmunity. Current limitations include sparse annotations, batch effects, incomplete structural coverage and limited model interpretability. Standardized datasets and interpretable multi-omics models will be essential for translating AI-driven glycoanalysis into pathogen surveillance and precision immunotherapy. In conclusion, this review summarizes how artificial intelligence (AI), machine learning (ML) and deep learning (DL) are being applied to glycomics and glycoanalysis in infection biology.