Figure 3

Experimental and randomized histograms for the functions macromolecule catabolism, response to desiccation, and biopolymer metabolism. A density measure that is evaluated for each data point is given by the number of neighbors that are close according to a product similarity measure. The blue bars in the histograms represent the sum of all these densities. Those vector pairs are expected to be most relevant that exhibit a large absolute value of differential expression and a small angle between vectors. The white bars show the randomly distributed vectors that are expected to have some neighbors as well. Subsets of genes that have the same number of elements as the protein function under consideration were randomly selected. The histogram was constructed for this random subset. The process was repeated multiple times and averaged over 20 runs. The top panel shows the histogram for the macromolecule catabolism function, the middle panels for the response to desiccation function, and the bottom panel for the biopolymer metabolism function.