Structure-Based and Glycosylation-Aware Interpretation of Glycogene Variants

Abstract Missense variants in glycan-related genes (glycogenes) underlie many congenital disorders of glycosylation and related conditions, yet a large fraction remain classified as variants of uncertain significance. Because glycogenes often show low evolutionary conservation, conventional sequence-based predictors are frequently uninformative, making structure-based interpretation particularly valuable. This review examines how structure-based and glycosylation-aware approaches can improve the interpretation of glycogene variants. Advances in protein structure prediction, notably AlphaFold, now allow variant-induced structural perturbations to be assessed proteome-wide using structure-derived features such as solvent accessibility, thermodynamic stability (ΔΔG), and local model confidence (pLDDT). Such structural features can be combined to predict the impact of glycogene variants, as exemplified by VarMeter and VarMeter2, which have successfully identified pathogenic missense variants of arylsulfatase L (ARSL) and N-sulfoglucosamine sulfohydrolase (SGSH) in patients with undiagnosed disease. Despite the progress in these tools, there remain challenges especially relevant to glycobiology, including the limited accuracy of current ΔΔG predictors, the distinct behavior of gain-of-function and stabilizing variants, the need for proteome-scale computation, and the critical but still underexplored contribution of protein glycosylation to structural stability. We argue that incorporating glycosylation into structural energetics is a key next step toward more accurate, glyco-aware variant interpretation.