Figure 4

Two example domain sequence clusters and the associated protein function annotations. All domains are coloured and labelled according to their true functions; the available high-quality GO annotations of the parent proteins are shown on the right, respectively. The GO terms are coloured according to the protein functions they describe, and their hierarchical relationships in the GO DAG are shown at the bottom, respectively (dashed lines represent omitted intermediate terms). Both clusters (dashed boxes) represent functional domain families according to criteria outlined in the main text. The three reductase functions in (a) are closely related, as indicated by the three-functional cluster member sequences. In (b), the hydrolase function is perfectly conserved among all member domain sequences. Note that both the true domain functions and the types of the available annotations (core, extra and foreign-domain; see main text) are 'invisible' to the core set identification protocol. The small numbers next to specific functions indicate in which step of the iterative protocol described in the main text they are identified as 'core-associated' functions (see annotation editing section).