Figure 6
From: N-glycosylation in non-invasive and invasive intraductal papillary mucinous neoplasm
Multivariate (OPLS-DA) modelling of healthy controls, non-invasive IPMN and invasive IPMN samples by neutral and acidic glycans. Score plot shows how well the predictive component (t1) of multivariate model can separate the samples from each other. (A) Scoreplot of healthy pancreatic tissue vs. non-invasive IPMN (the predictive component (t1) on the x-axis and the orthogonal component (to1)on the y-axis, permutation test for Q2Y p = 0.002 and for R2Y p = 0.002), (B) Score plot of healthy pancreatic tissue vs. invasive IPMN (permutation test for Q2Y p = 0.001 and for R2Y p = 0.001), C) Score plot of non-invasive IPMN vs. invasive IPMN (permutation test for Q2Y p = 0.111 and for R2Y p = 0.101). R2X and R2Y are proportion of predictor/response variation explained by the full model, respectively. Q2Y is predictive performance of the model, RMSEE is root mean squared error of estimation, pre is the number of predictive components, ort is number of orthogonal components. Both neutral and acidic proposed monosaccharide compositions with the Mann–Whitney U test p-values of less than 0.1 were used in the models. Black ellipse is the 95% confidence interval for the model and colored (green = healthy, blue = non-invasive and red = invasive) ellipses are the 95% of multivariate normal distribution for each group of samples. The percentage of response variance explained by the predictor component (t1) is indicated in parentheses.
