Skip to main content

Table 1 Accuracy of the eight scRNAseq data classifier tools and the four conventional classifiers on the seven datasets (N = 1000)

From: Single-cell classification using graph convolutional networks

Methods Zhengsorted Zheng68K BaronHuman Muraro Segerstolpe BaronMouse Xin
sigGCN 0.922 0.752 0.979 0.991 0.977 0.974 0.993
FC 0.893 0.668 0.967 0.986 0.967 0.958 0.993
scID 0.721 0.484 0.46 0.577 0.285 0.286 0.986
scPred 0.515 0.140 0.86 0.915 0.827 0.862 0.91
CasTLe 0.836 0.736 0.971 0.972 0.953 0.91 0.993
SingleR 0.723 0.388 0.951 0.977 0.953 0.868 1
scmapcluster 0.395 0.409 0.946 0.962 0.949 0.905 0.931
scmapcell 0.727 0.246 0.895 0.972 0.949 0.778 0.952
ACTINN 0.845 0.737 0.977 0.991 0.958 0.984 1
RF 0.835 0.69 0.968 0.981 0.967 0.968 0.993
SVM-linear 0.859 0.652 0.824 0.981 0.383 0.704 1
SVM-rbf 0.884 0.677 0.93 0.981 0.841 0.794 1
KNN 0.824 0.594 0.953 0.991 0.935 0.926 1