Fig. 3.

Simulated annotation database. (a) Ideal annotation database: annotated gene category contains exactly the same genes as embedded module. This is the case when we have ideal annotation databases. (b) Incomplete database: only parts of genes in an embedded module are annotated. This is the case when not all genes are annotated. (c) Incomplete and noised database: annotated gene category and the corresponding embedded module have overlap. It implies that not all genes are annotated; some annotations might be inaccurate or unsuitable to the specific data. This is a common case.