Figure 5From: EVEREST: automatic identification and classification of protein domains in all protein sequencesPerformance of EVEREST release 1, ProtoNet and ADDA with respect to Pfam and SCOP. A. Accuracy with respect to Pfam. Histogram of scores of EVEREST/ProtoNet/ADDA families with respect to best matching Pfam families. B. Coverage of non-trivial Pfam families: Histogram over non-trivial Pfam families, of scores of best matching EVEREST/ProtoNet/ADDA families. C. Coverage of hetero Pfam families: Histogram over hetero Pfam families, of scores of best matching EVEREST/ProtoNet/ADDA families. D, E, F: As A, B, C, with SCOP in place of Pfam. G. EVEREST Accuracy w.r.t Pfam: Two-dimensional histogram of sensitivity and selectivity of EVEREST families with respect to best matching Pfam families. H. EVEREST Coverage of hetero Pfam: Two-dimensional histogram over hetero Pfam families, of sensitivity and selectivity of best matching EVEREST families. I, J: As G, H, with SCOP in place of Pfam. Bins count values above lower threshold, and below or equal to upper threshold (for coverage, Pfam/SCOP families that are not intersected by any EVEREST (ProtoNet/ADDA) family are added to the lowest bin). In G, H, I and J, numbers are percentages of families in bin. See sections Evaluation of Classification and Evaluating a Suggested Domain Family for further details.Back to article page