Data Science in Glycoscience

Abstract

Glycans contribute in numerous biological processes, including cell-cell communication, immune recognition, pathogen interaction, and protein folding. These functions largely arise through glycosylation, one of the most prevalent post-translational modifications of proteins, which strongly influences the stability, localization, and activity of biomolecules. Beyond their biological significance, carbohydrates also represent important targets in synthetic chemistry, where the controlled formation of glycosidic bonds remains a central challenge. In particular, the stereoselective construction of glycosidic linkages is highly sensitive to multiple interacting factors, including sugar structure, protecting-group patterns, sugar reactivity, solvent, temperature, concentration, and promoter systems. The inherent structural complexity of glycans, characterized by branching architectures, multiple linkage positions, and stereochemical variability—creates a vast chemical space in which reaction outcomes are often difficult to predict using mechanistic intuition alone. Unlike nucleic acids and proteins, glycans lack a direct template-based encoding, further contributing to structural heterogeneity and complicating systematic exploration of glycosylation chemistry. Recent advances in data science and artificial intelligence provide new opportunities to analyze complex glycoscience datasets and uncover quantitative relationships between molecular structure, reaction conditions, and glycosylation outcomes. This chapter examines how data-driven approaches are beginning to address longstanding challenges in glycosylation chemistry by enabling quantitative analysis of complex reaction datasets and revealing relationships between molecular structure, reaction conditions, and stereochemical outcomes. Collectively, these developments illustrate how data-driven approaches can complement mechanistic insight and contribute to a more predictive framework for glycosylation chemistry.