Figure 3
From: Conceptualization of molecular findings by mining gene annotations
Example of distributions of graph-based statistics and the discriminative power of coherence models. A. In the scatter plot (top left), statistics derived from KEGG gene sets (red), the matching random gene sets (blue), and simulated random gene sets during model-building (green) were plotted. Top right is the corresponding ROC curve. B. Scatter plot of the graph-based statistics and ROC curve of the model using IC as information loss and Steiner tree length derived from augmented GO graph.