Skip to main content
letter
. 2021 Jul 7;12:4168. doi: 10.1038/s41467-021-24304-4

Fig. 2. Comparison of the performance of ImmuneCells.Sig with other ICT (immune checkpoint therapy) response signatures and independent validation of ImmuneCells.Sig without or with batch effect correction.

Fig. 2

The multiple bar plots of the fivefold cross-validation calculated mean testing AUC (Area Under The Curve) values of the whole 13 ICT signatures are shown in a For the GSE78220 data set. b For the GSE91061 data set. c For the PRJEB23709 data set. d For the MGSP data set. e Testing the predictivity of the ImmuneCells.Sig trained in the GSE91061 data set in the other three independent data sets—PRJEB23709, GSE78220, and MGSP without or with batch effect correction. f Testing the predictivity of the ImmuneCells.Sig trained in the PRJEB23709 data set in the other three independent data sets—GSE78220, MGSP, and GSE91061 without or with batch effect correction. g Testing the predictivity of the ImmuneCells.Sig trained in the GSE78220 data set in the other three independent data sets—MGSP, PRJEB23709, and GSE91061 without or with batch effect correction.