Skip to main content
Figure 2 | BMC Bioinformatics

Figure 2

From: ROC and confusion analysis of structure comparison methods identify the main causes of divergence from manual protein classification

Figure 2

Confusion matrix heat map. Confusion matrix heat map for VAST with a Pcli cutoff value of 2.5 and for SHEBA with a Zscore cutoff value of 2.7. The cutoffs correspond to an overall average FPR of 0.01, and result in an overall average TPR of 0.616 and 0.748 for VAST and SHEBA respectively. The x (target folds) and y (query folds) axes of the heat maps are labeled by the SCOP folds, grouped into classes A, B, C, D, E, F and G. Each class is delimited by a vertical line (for the x axis) and a horizontal line (for the y axis). Each pixel within the heat maps represents a fold-specific true or false positive rate and takes value between 0 and 1. Diagonal and off-diagonal pixels correspond to fold-specific true positive rate TPR i (c) (eq. 4, see Methods) and fold-specific false positive rate FPRi,j(c) (eq. 3, see methods) respectively. To improve the visibility of the heat maps, rates between 0 and 0.2 are represented in grey scale where white corresponds to a rate of 0 and black to a rate at or above 0.2. For high resolution heat maps of VAST and SHEBA, [See Additional file 1].

Back to article page