Additional file 5.

Figure S2. Detailed performance comparison under all the 20 thresholds. Figure A to Figure T shows the comparison result of two methods to identify CCRGs under 20 sets of thresholds. A is the result under the following threshold: degree_threshold: percentile 1 (0.01), betweenness centrality_threshold: percentile 1(0.01). B is the result under the threshold: degree_threshold: percentile 2 (0.02), betweenness centrality_threshold: percentile 2 (0.02). And the corollary, Figure T is the result under the threshold: degree_threshold: percentile 20 (0.20), betweenness centrality_threshold: percentile 20 (0.20). The text over each figure is the area under curve (AUC). Take Figure A for example, 0.7087 vs 0.5446 represents that AUC of our method is 0.7087, and 0.5446 is AUC of traditional method based on gene expression. The AUC is colored according to curve color.

Format: TIFF Size: 1MB Download file

Chen et al. BMC Medical Genomics 2012 5:43   doi:10.1186/1755-8794-5-43