VarMeter: a prediction method for the impact of glycogene variants

The clinical relevance of glycans, which play a wide array of physiological roles, is underscored by the emergence of congenital disorders of glycosylation, a group of rare inherited diseases caused by defects in glycan-related genes (glycogenes). Biochemical studies of recombinant proteins and phenotypic analyses in knockout mice are revealing critical insights into the roles of various glycosyltransferases, glycosidases, and glycan-binding proteins. However, the biological functions of numerous glycogenes and their role in disease remain incompletely understood, partly due to human-specific functions that are not recapitulated in model organisms, and partly due to the structural diversity and complexity of glycan modifications, which are difficult to fully assess by conventional methods. A promising complementary strategy is the systematic assessment of human genetic variants, particularly missense mutations, to infer functional consequences. Recent developments in protein structure prediction, exemplified by AlphaFold, are facilitating the development of structure-based approaches to variant interpretation. In this review, we discuss current methodologies for predicting the impact of missense variants using structural information, and introduce VarMeter, a computational framework incorporating 3D structural parameters that has been successfully applied to the prediction of pathogenic variants in the ClinVar database. We also describe VarMeter2, an updated version that integrates AlphaFold-derived pLDDT confidence scores and Mahalanobis distance analysis to improve prediction accuracy, demonstrating its ability to predict pathogenic variants of four glycan-related proteins. These tools offer a novel avenue for uncovering previously unrecognized functions of glycogenes and their links to disease, and contribute to the clinical interpretation of genetic variation.