Hi all, I run an OPLS-DA and worked fine. The problem I have is that the output of permuted model is identical to the original model and shows the the predictive component significant. I didn't understand how the permuted model performs identically to the original model and at the same time the pQ2Y=0.002. The score plot is also identical. Thank you

dataMatrixEF<-EFFECT.data[,-c(229:230)]
EF_oplsda <- opls(dataMatrixEF, Y_EF, predI = 1, scale="standard", orthoI = NA, subset=c(1:100))
               R2X(cum) R2Y(cum) Q2(cum) RMSEE RMSEP pre ort
Total             0.68         0.813         0.57      0.224    0.318   1   5
Warning message:
'permI' set to 0 because train/test partition is selected

EFperm<-dataMatrixEF[c(1:100),]
EFperm_oplsda <- opls(EFperm, perm_Y_EF, scaleC="standard",predI = 1, orthoI = NA,permI=500)

          R2X(cum) R2Y(cum) Q2(cum) RMSEE pre ort  pR2Y   pQ2
Total     0.68            0.813          0.57      0.224   1   5     0.002 0.002

Hi, If the permutated response is identical to the intial vector, the corresponding model is identical. This is why the similarity between the permutated and original response is plotted in abscissa in the "permutation" plot assessing the significance of the modeling.