Figure 2.

Predictive ability of PLS discriminant analysis (PLS-DA) models estimated by five fold-cross-validation. Kinase descriptions in the six models presented in panels A-F correspond to respective panels in Figure 1. Q² values may vary from 0 to 1. A low value indicates that a kinase group is randomly mixed with the others, while a value approaching 1 indicates a perfect discrimination. The overall high Q² values certify that the respective types of kinase descriptions have captured sequence properties that are present in members of a specific kinase group while being absent among all other kinases.