Fig. 3.

Comparison of phenotype frequencies expected from a computational gold standard versus experimentally validated frequency. Expected mitochondrial phenotype area under the precision/recall curves (AUPRCs) were calculated from Figure 2A using only the GO; experimentally validated AUPRCs use the augmented standard of Figure 2D. Phenotype frequencies and accuracy of computational predictions are much higher in all cases than anticipated by pre-existing functional catalogs.