Figure 4
From: Integrating phenotype and gene expression data for predicting gene function
Total Number of Positive Predictions. As the cutoff used in the prediction algorithm is increased, all of the datasets make fewer positive predictions. That is, they predict that fewer genes should be annotated with a particular GO term. However, the number of predictions based on only the phenotype data is consistently far less than the number based on the expression data or the combined data set. This suggests that, in general, the phenotype data will not be as much aid in making novel predictions as the other data sets. There is no difference in the number of predictions assigned by either the generalized or exact approaches since those only differentiate between which predictions are considered correct.