The results of DCN-based aggregated neighbor-counting method on 66 randomly selected proteins.

The ways of calculating precision and recall can be found at the captions of Table 4 and Table 3, respectively. Explanations of the best semantic similarity score can be found at the caption of Table 2. From the 100 proteins randomly selected from GO database, 66 proteins have predictions available by DCN-based aggregated neighbor-counting method.