Figure 8

Comparison of the AUCs (Area Under the Curve) between our proposed kernelized score functions and othe machine learning methods for gene ranking, using FI , HumanNet and integrated networks. AUCs are averaged across all the modules and estimated through 5-fold cross-validation repeated 5 times. S _AV (Average score), S _NN (Nearest-neighbor score) and S _kNN (k-Nearest-neighbor score) represent kernelized score functions. The parameter k of S _kNN is set to 27. Zhou is the algorithm based on Gaussian Random Fields proposed in [31] and GeneMANIA its variant, while LabelProp is the Label Propagation algorithm proposed in [17].