Skip to main content
. 2008 Oct 23;9:452. doi: 10.1186/1471-2105-9-452

Figure 4.

Figure 4

Performance of GCRMA-R, GCRMA-L and GCRMA-NN on the Latin square dataset. A) ROC curves showing the average true positives and false positives across the 14 2× Latin square experiments following application of the cyber t test. B) The number of true positives captured for all the 14 2× Latin square experiments at a cutoff value of four false positives (dashed vertical line in panel A). The differences in panel B between GCRMA-R, GCRMA-L and GCRMA-NN are statistically significant (p < 0.005) using paired one-sided Wilcoxon and t tests.