Figure 4From: EnzML: multi-label prediction of enzyme classes using InterPro signaturesCross-evaluation results. The plot compares the subset accuracy between taxonomic datasets and random sets of the same size. The rightmost point of the diagram is the whole SwissProt⋈KEGG dataset. The y axis (accuracy and recall) starts at 70%. An asterisk indicates significant difference in accuracy (with p-value at 5%) between the taxonomic and random datasets below. The full data is available in Additional file 7: all_cross_evaluation_results.csvBack to article page