Figure 1

A schematic diagram of the five-fold cross-validation scheme for the training and testing of the SVRMHC model constructed for H2-Kk (154 peptides), with enclosing parameter searching modules in which leave-one-out (LOO) cross-validation was used. The models for the other two datasets (for H2-Db and H2-Kb) were constructed similarly, with the exception that the computationally more expensive LOO cross-validation (rather than five-fold cross-validation) was used on the outer-loop model training and testing procedure.