Fig. 3: N-glycosylation in a structural context.

a, Glycosylated asparagines highlighted as spheres in the AlphaFold-predicted structure of the coxsackievirus and adenovirus receptor (P78310). Colors indicate the model predicted local distance difference test (pLDDT). b, Total number of identified glycans and relative glycan type composition per site in the same protein. Topological domains and IDRs are also highlighted (IDR definition in Methods). c, Validation AUROC for the predictive model of lowly and highly glycosylated asparagines that uses structural features as input (details in Methods; low-N = 2,203 and high-N = 747 in mouse; low-N = 1,196 and high-N = 488 in human). d, Feature importance plot. The x axis indicates the F score for every feature on the y axis. e, pPSE (24 Å, 180°) for highly and lowly glycosylated asparagines (low-N = 2,203 and high-N = 747 in mouse; low-N = 1,196 and high-N = 488 in human). f, Log-transformed number of glycoforms on asparagines within extracellular (N = 2,813) or cytoplasmic (N = 335) topological domains. g, Average AlphaMissense score for lowly (N = 1,196) and highly (N = 488) glycosylated asparagines on data from human cell lines. All box plots indicate the median and the first and third quartiles. Whiskers extend from the hinges to the largest value no further than 1.5 × the interquartile range. Data points beyond the end of the whiskers are plotted individually. All P values indicate the comparison of distributions using a two-sided Wilcoxon rank-sum test.