Multi-labelling evaluation—non-perfect label match.
收藏资源简介:
Illustrations:[1]True Subcellular Locations : denotes the labels from Hum-mPLoc 2.0 dataset (GOA database version 70.0 released March 10 2008);[2]Cytoskeleton[0.136]:the label NOT included in the original Hum-mPLoc 2.0 dataset, but validated TRUE by the latest Swiss-Prot database (http://www.uniprot.org/uniprot/UniProt release 2011_11 Nov 16, 2011), where [0.136] denotes the probability that the protein is assigned to the label Cytoskeleton;[3]Nucleus[0.115]: Non-target Label Hit (wrong prediction), NOT validated by the latest Swiss-Prot database (http://www.uniprot.org/uniprot/UniProt release 2011_11 Nov 16, 2011), where [0.115] denotes the probability that the protein is assigned to the label Nucleus;
标注说明:[1] 真实亚细胞定位(True Subcellular Locations):指源自Hum-mPLoc 2.0数据集(GOA数据库70.0版本,发布于2008年3月10日)的标签;[2] 细胞骨架(Cytoskeleton)[0.136]:该标签未被原始Hum-mPLoc 2.0数据集收录,但经最新版Swiss-Prot数据库(UniProt 2011_11版本,发布于2011年11月16日,访问地址:http://www.uniprot.org/uniprot/)验证为真实标签,其中[0.136]代表该蛋白被标注为细胞骨架标签的概率;[3] 细胞核(Nucleus)[0.115]:非目标标签命中(即预测错误),未通过最新版Swiss-Prot数据库(UniProt 2011_11版本,发布于2011年11月16日,访问地址:http://www.uniprot.org/uniprot/)的验证,其中[0.115]代表该蛋白被标注为细胞核标签的概率;



