Fig. 7: The EV N-glycopeptide signatures of seminal plasma show AZS subtypes.

A Consensus clustering analysis categorized AZS samples in the discovery cohort into two subtypes: AZS-C1 (n = 9) and AZS-C2 (n = 11). B Distribution of samples from AZS subtypes, AZS-C1 (n = 9) (red) and AZS-C2 (n = 11) (blue), in the PCA plot of the discovery cohort. C Heatmap displays differential expression of EV N-glycoproteins between AZS subtypes AZS-C1 and AZS-C2 (P-value < 0.05, two-sided independent Student’s t-test). The color range, from red ( + 2) to blue (-2), indicates the row z-score of normalized N-glycoproteins expression levels. D SHAP values for the ten important features (N-glycoproteins) distinguishing AZS-C1 (n = 9) and AZS-C2 (n = 11) subtypes in the discovery cohort. E The ROC curves of the ten important N-glycoproteins used to distinguish AZS subtypes, AZS-C1 (n = 9) and AZS-C2 (n = 11), in the discovery cohort. F, G Abundance distributions of the N-glycoproteins RAB4B (F) and CXADR (G) in the discovery cohort. Error bar, median with interquartile range. AZS-C1 (n = 9) (sky blue), AZS-C2 (n = 11) (scarlet). The P-value was calculated using two-sided independent Student’s t-test. H AZS samples in the verification cohort were also categorized into two subtypes AZS-C1 and AZS-C2 by K-means clustering analysis. (I) The ROC curves to distinguish AZS-C1 (red) and AZS-C2 (cyan) in the verification cohort. (J) Abundance distributions of N-glycoproteins PLA1A in the verification cohort. AZS-C1 (n = 12) (sky blue), AZS-C2 (n = 22) (scarlet). The P-value was calculated using two-sided independent Student’s t-test. Error bar, median with interquartile range. Source data are provided as a Source Data file.