Figure 6
From: EnzML: multi-label prediction of enzyme classes using InterPro signatures
Cross-evaluation on the UniRef reference sequences. Left panel: the reference sequences are derived from SwissProtKEGG using UniRef100, UniRef90 or UniRef50 clusters. Right panel: number of protein instances, InterPro attributes and EC classes when the SwissProtKEGG dataset is reduced to its UniRef representative sequences. The values in both panels are shown as difference to the corresponding value for the entire SwissProtKEGG dataset. The full data is available in Additional file 7: all_cross_evaluation_results.csv