Fig. 3

Visualization of multi-label genomic data. PROMO provides a variety of methods for visualizing a genomic dataset together with its associated clinical information. a A multi-label expression matrix plot. The plot is composed of a heat-map representation of the genomic matrix and several label bars beneath it showing different clinical labels that the user interactively selected. The colors in each label bar show the label value of each sample according to the legend on the right. The label appears below the lower left corner of the bar. Here, breast cancer patient profiles were grouped according to their PAM50 category (shown in the top label bar). By observing the distribution of values in other bars, relations between the groups and the labels can be observed. For example, the ER, PR and HER2 status of most samples in the ‘Basal’ group are negative, while the HER2 status of most ‘HER2’ group is positive. b Data distribution and c Clinical label distribution can be explored and visualized separately, or in combination using plots such as d PCA and others. These figures show that the Basal tumor samples are mainly characterized by Negative ER, PR and HER2 labels (a) and markedly differ from all other subtypes in their gene expression pattern (d), in accordance with the literature [14]