Skip to main content
. 2015 Mar 26;43(Web Server issue):W122–W127. doi: 10.1093/nar/gkv253

Figure 1.

Figure 1.

Network assessment using a set of validation gene pairs based on agriGO biological process (BP) annotations. (a) Precision-recall analysis to compare RiceNet v1 and RiceNet v2. The precision of co-functional links were measured by odds ratio (OR). RiceNet v2 shows substantially higher OR than RiceNet v1 for high scored network links. (b) A box-and-whisker plot of network prediction powers for 336 agriGO BP terms with more than four annotated genes, measured by area under the curve from ROC analysis. RiceNet v2 shows significantly higher prediction power for the processes than RiceNet v1 (P = 1.11 × 10-6, Wilcoxon signed rank test). To avoid network size effect on prediction power, we used only top 588 221 links of RiceNet v2 to match the size of RiceNet v1 during analysis.