Hierarchical clustering of cellular compartments based on selective constraint. Genes are split into nine groups based on their subcellular location (see main text). The median box plots show the distribution of ω values for each compartment - all data, including outliers, are used in the analysis. Using the mean ω values, we created a phenogram using the UPGMA algorithm (branch lengths are arbitrary). Each branch is colored gold if a Mann-Whitney U-test found that the distributions were significantly different at P ≤ 0.01, and blue otherwise. For example, the cytoplasm and mitochondria distributions are significantly different (P = 0.01), but the lysozyme and transport groups show no significant difference (P = 0.087). However, when the group formed by the cytoplasm and mitochondria is compared to that formed by the lysozyme and transporters, there is a statistically significant difference (P < 0.001). The 4 Venn diagrams show the proportional degree of overlap in genes among groups (sizes are not comparable across nodes in the tree). In none of these cases is any one set of reactions a superset of the other set of reactions.
Hudson and Conant BMC Evolutionary Biology 2011 11:89 doi:10.1186/1471-2148-11-89