Figure 4. A comparison of the distribution of GO annotations in the synthetic datasets generated using the three approaches and the distribution in the target dataset in the three sub-ontologies: (a): Cellular Component, (b) Biological Process, (c) Molecular Function.