Figure 9
From: Integrating phenotype and gene expression data for predicting gene function
F-measures at different depths within the GO. Because of the more specific information available in the phenotype data set, it results in more accurate predictions at deeper levels in the GO. This figure shows that for predictions at levels between 7 and 12 in the GO, the phenotype data almost always has a higher F-measure. However, by combining the phenotype data with the expression data, the many more predictions (see Figures 4 and 5) made by the integrated data set do fare better than those of the gene expression data set alone. These were only evaluated using the generalized predictions. A cutoff of 0.6 was used in all cases.