Extended Data Fig. 10: Overlap of results with proteome-wide T2D associations in AGES-Reykjavik.

a) Contingency table tabulating the overlap in results from our study detailed in Extended Data Fig. 3 (rows) with proteome-wide significant associations with incident and prevalent T2D in AGES-Reykjavik in Gudmundsdottir et al. 2020³⁵ (columns). One-sided P-values from Fisher’s exact tests are given in each cell testing whether the overlap is greater than expected by chance. Row totals and column totals indicate the number of proteins in each row and column group, and the total overlap in proteins present in both studies (3,250) is given in the bottom right. b) For the 16 of 31 proteins nominally associated with T2D PGS in INTERVAL (Fig. 2b) and proteome-wide significant for incident T2D in AGES-Reykjavik, compares hazard ratios (points; x-axis) for incident T2D in INTERVAL (N = 27 cases over 7.7 years of follow-up in 3,087 participants) to odds ratios (points; y-axis) for incident T2D in AGES-Reykjavik (N = 112 cases after 5 years of follow-up in 2,940 participants). Cox proportional hazards models in INTERVAL were fit with follow-up as time scale, adjusting for age, sex, 10 genotype PCs, sample measurement batch, and time between blood draw and sample processing. Logistic regression in AGES-Reykjavik were fit adjusting for age and sex³⁵. Horizontal and vertical bars correspond to the 95% confidence intervals of the hazard ratios and odds ratios respectively. Two-sided P < 0.0012 indicates association with incident T2D in INTERVAL from Fig. 2b was significant after Bonferroni correction for the 42 tested protein to disease associations. Summary statistics including exact two-sided P-values from both analyses are given in Supplementary Table 8.