Figure 6

Intermediate analysis steps in Furby. (a) The analyst starts by inspecting the overview visualization showing the biclustering result of the multiple tissue types dataset. The initially set membership value thresholds result in a large number of overlaps in both dimensions. Consequently, also the visual representation in Furby is very cluttered. By optimizing the global membership thresholds, the analyst gets a much cleaner representation, as shown in Figure 3. In (b), the analyst inspects a bicluster with high membership values in gene and sample dimension. The additional metadata bar in sample dimension encodes the tissue type (purple = prostate). The metadata bar in gene dimension represents the 4 top results of a DAVID functional annotation analysis (green = part of gene set, black = not part of gene set). The cluster in (c) contains lung (green) and breast tissue samples (yellow) that show an inverse gene regulation.