Figure 1
Univariate results (A) Analyses (‘volcano plots’) of benign compared to early stage (stage I and II) in the discovery cohort. Mean NPX differences (malign–benign) are shown on the x-axis and p-values (− log10, two-sided Wilcoxon ranked test) on the y-axis. Proteins plotted with a ‘ + ’ were significantly different in the discovery data (q < 0.05, Holm-adjusted). Proteins plotted in red were also found to be significant (q < 0.05, Holm-adjusted) also in the replication data while proteins plotted in black were not. Up to 10 proteins with the lowest p-values in the discovery data and/or the largest NPX-difference are labelled. (B) Same as (A) but for benign compared to late stage (stage III and IV). (C) Same as (A) but for benign compared to any stage (stage I–IV). (D) Beenplots of individual protein measurements for RBFOX3 in the discovery cohort (left side) and replication cohort (right side). Group mean is indicated with a black line and individual measures with thin blue lines. The samples are divided by diagnose: B—benign (coloured grey), I, II, III and IV—ovarian cancer FIGO stage (coloured yellow to red). (E) As (D) but for TCOF1. (F) ROC-curve for RBFOX3 (solid) and MUCIN-16 (dashed) in the replication cohort using a model developed in the discovery cohort. The AUC is indicated with 95% confidence interval. (G) As (F) but for TCOF1 and MUCIN-16.
