Additional file 7 of Assignment of structural domains in proteins using diffusion kernels on graphs

Additional file 7. KluDo's performance based on the number of domains over ASTRAL40. This file consists of the accuracies over the results in Additional file 4 separated by the number of domains. Based on each of the SCOP and CATH databases, four subsets were extracted from ASTRAL40: 1-domain, 2-domain, 3-domain, and 4-domain structures. For each subset, the percent of the correct assignments (the cases of compliance with SCOP or CATH based on the OL score using an 85% threshold), overcuts (the cases of assigning a higher number of domains than both SCOP and CATH), undercuts (the cases of fewer domains than both SCOP and CATH), boundary inconsistencies (the cases of incorrect assignment where the number of domains complies with SCOP or CATH) and other cases were measured. CA, OC, UC, and BI represent correct assignments, overcuts, undercuts and boundary inconsistencies, respectively. Also, KK and SP stand for kernel k-means and spectral clustering, respectively.

补充材料7：KluDo在ASTRAL40数据集上基于结构域数量的性能表现
本文件包含按结构域数量分组的补充材料4中的精度统计结果。分别以SCOP（蛋白质结构分类数据库，Structural Classification of Proteins）与CATH数据库为基准，从ASTRAL40数据集中提取得到四个子集：单结构域、双结构域、三结构域及四结构域蛋白结构样本。针对每个子集，本研究统计了以下几类结果的占比：正确分配结果（即基于OL得分（OL score）、以85%为阈值，匹配SCOP或CATH分类标准的分配情况）、超分配（即分配的结构域数量多于SCOP与CATH标注结果的情况）、欠分配（即分配的结构域数量少于SCOP与CATH标注结果的情况）、边界不一致（即结构域数量符合SCOP或CATH标准但分配边界存在错误的情况）及其他异常情况。其中CA、OC、UC及BI分别代表正确分配、超分配、欠分配与边界不一致。此外，KK与SP分别代表核k均值（kernel k-means）与谱聚类（spectral clustering）。