Fig. 4 | Scientific Reports

Fig. 4

From: Genome-wide investigation of gene-cancer associations for the prediction of novel therapeutic targets in oncology

Fig. 4

Generalization performances on the test sets for the best models across cancer types, measured in terms of AUROC. Left: Models trained on the three “primary” features (mutation, expression, essentiality). Right: Models trained after integrating the network embedding features and performing feature selection. The dashed lines correspond to a theoretical null model that predicts a label at random (AUROC = 0.5).

Back to article page