Figure 3

Model building procedure published gene sets. Model building procedure for testing the performance of published gene sets that predict ovarian cancer outcomes. First (A), the gene set is extracted from the literature and the corresponding genes are selected in both the pilot and independent data set. Subsequently, (B) an LS-SVM model is built using only the pilot data and finally, (C) this model is used to predict the class of the samples in the independent data set which gives an estimate of the generalization performance of a gene set.