Fig. 1

Principal component analysis (PCA) of 132 metabolite concentrations in the plasma of ponies after an overnight fast (basal), or feeding a high-NSC diet (postprandial) from a cohort that contained a group of animals that subsequently developed laminitis (Pre-laminitic, red), and a group that did not (Non-laminitic, blue). To represent high dimensional data, PCA creates a linear combination of the dataset features such as to capture the maximum variance in the data. The amount of total variation explained by the two first principal components is given along the x and y axis, respectively. Each point represents one plasma sample from one pony either in the basal (triangle) or postprandial state (circle). The ellipses represent the 68% confidence interval of each group. The basal laminitic and non-laminitic samples (a) fell into two clusters which were best separated by features associated with PC2. In contrast, there was a greater overlap between the groups for the postprandial samples (b). When the factors of group and feeding were combined (c), some separation between groups was still evident for the basal samples, whereas the groups clustered together in the postprandial samples. A good separation between clusters indicates that it is possible to distinguish both groups by a linear combination of features