Figure 2

Validation of predictions using known annotations in diverse subcategories in GO. The prediction performance is shown here on a total of 27 GO subcategories: 14 for BP, eight for MF, and five for CC subcategories (see Methods). In each panel, the title is the name of GO subcategory with a label prefix "P", "F", or "C" to denote biological process, molecular function, or cellular component, respectively. The number in parentheses indicates the total number of ORFs annotated in each GO subcategory. A plot of precision (y-axis) as a function of sensitivity ratio is shown for all unknown ORFs (x-axis), as defined in Methods. In such plots, the upper right corner corresponds to a perfect predictor because of the larger area under the precision–sensitivity curve.