Class prediction of 144 transcripts from the Training Set.

Class prediction was performed using support vector machines (SVM). The 144 transcripts derived from the Training Set were used to build the SVM model, the model was then run in the other two cohorts and the external cohort (Maertzdorf et al).