Figure 1.

Circular component. The gene expression data of the IDC form a circular structure caused by variation over time. Circular PCA is used to identify a circular component (red line) that approximates the data and, hence, provides a noise-reduced and continuous model of the IDC. The component describes a curve located in the high-dimensional data space given by all genes. The component and the original data are visualized by plotting them into the reduced space given by the first three (linear) principal components (PC1-3) of standard PCA.