Figure 5From: Learning virulent proteins from integrated query networksImprovement over naïve prediction. Improvement in AUC of using an SVM for classification for cross-validated tests over Top−F 1. The colors in each cell correspond to the difference in AUC between using an SVM for prediction and using the top-performing feature per source, per label. Higher values indicate cases where there is added benefit in considering multiple features via SVMs.Back to article page