Prediction analysis. PAM using 26 randomly selected samples as a training set and the remaining 25 samples as a test set. The probability of a true prediction and the prediction results are shown. If the prediction probability was less than 0.5, the sample prediction was wrong (□). (A) Samples were prepared as the training samples (test set 1). (B) RNA preparation of samples was different, as described in “Patients, materials, and methods” (test set 2).